Professional Documents
Culture Documents
Hans-Peter Degischer,
Brigitte Kriszt (Editors)
Handbook of Cellular Metals
Production, Processing,
Applications
Continuous Casting
ISBN 3-527-30283-2
G. Santhoff
Intermetallics
ISBN 3-527-29320-5
Edited by
Hans-Peter Degischer, Brigitte Kriszt
Editors
Prof. Dr. Hans-Peter Degischer
Technische Universitat Wien
Institut fur Werkstoffkunde
und Materialprufung
Karlsplatz 13
A-1040 Wien
Austria
Dr. Brigitte Kriszt
Technische Universitat Wien
Institut fur Werkstoffkunde
und Materialprufung
Karlsplatz 13
A-1040 Wien
Austria
This book was carefully produced. Nevertheless, authors and publisher do not
warrant the information contained therein
to be free of errors. Readers are advised to
keep in mind that statements, data, illustrations, procedural details or other items
may inadvertently be inaccurate.
Library of Congress Card No.: applied for
British Library Cataloguing-in-Publication
Data:
A catalogue record for this book is available
from the British Library.
Die Deutsche Bibliothek
CIP-Cataloguing-in-Publication-Data
A catalogue record for this book is available
from Die Deutsche Bibliothek
c WILEY-VCH Verlag GmbH
69469 Weinheim, 2002
All rights reserved (including those of
translation in other languages). No part of
this book may be reproduced in any form
by photoprinting, microfilm, or any other
means nor transmitted or translated into
machine language without written permission from the publishers. Registered names,
trademarks, etc. used in this book, even
when not specifically marked as such, are
not to be considered unprotected by law.
printed in the Federal Republic of Germany
printed on acid-free paper
Composition Hagedorn Kommunikation,
Viernheim
Printing Strauss Offsetdruck GmbH,
Morlenbach
Bookbinding J. Schaffer GmbH & Co. KG,
Grunstadt
ISBN
3-527-30339-1
Preface
B. Kriszt and H. P. Degischer
There have been a few international and national research and development programs on cellular metals in recent years: the US Multidisciplinary Research Initiative (MURI) on ultralight metal structures since 1996; a few European research
projects funded within the 4th and 5th EU Framework Program; and a focussed
research program funded by the German Research Council, which began in
1998. These cooperative activities add to the original research activities at the
Fraunhofergesellschaft IFAM in Bremen, the University of Cambridge, Ranshofen
and Vienna in Austria, the Slovac Academy of Science in Bratislava, and several
other places in Europe. A considerable quantity of data has been produced and
presented at international conferences. One of these meetings was organized by
the German Society of Materials (DGM) on 28/29 February 2000 at the Vienna
University of Technology. The proceedings are mainly in German [1] and the
authors wanted another opportunity to publish their results in context with present
knowledge. The reader of proceedings might be overwhelmed by very specific research results on selected topics, obscuring a general understanding of this expanding field of research. Besides the classical books dealing with cellular metals
[2,3], an overview of the state of the art in the year 2000 was needed, covering primary and secondary processing, characterization of cellular metals, properties,
modeling, and exploitation.
As a result, some of the contributors to the Vienna symposium Metallschaume
2000 have been asked to extend their papers, by referring to related results of
other researchers and giving a review of their particular topic, whilst maintaining
the detailed specialist knowledge of the author. These contributions are introduced
by coordinators who describe the state of the art in that field. Foamed metals are
described more extensively than other cellular metals because of the actual research activities prevalent in Europe. The European development goal is the application of cellular metals in components for motor vehicles, a field in which price
limits and consistent high quality are essential. This handbook aims to give a more
detailed overview of the present state of the art in research and development on
cellular metals with specific emphasis on processing and characterization of
foamed metals than the recent survey [4]. The handbook gives a starting point
VI
Preface
for researchers new to the field, and references to topics adjacent to their own specialty for experts already engaged. Engineers and potential users are encouraged to
consider the application of cellular metals, taking into account the specific peculiarities of each material to avoid failures due to miscalculation of processing requirements and performance. A guide to the multitude of different types of cellular metals is provided with an indication of particular differences in properties. Research
is ongoing and it is to be hoped that experience with applications, that the book
intends to promote and stimulate, is expanding.
The editors thank all contributors and acknowledge their assistance in adjusting
the content of their contributions in cooperation with each other to produce a concise overview of the state of the art in cellular metals from a European viewpoint.
References
Foreword
T. W. Clyne
Interest in metallic foams dates from the 1940s, when Sosnick filed a patent on a
production method involving vaporization of low melting point constituents of metallic alloys [1]. Other publications and patents followed in the subsequent two decades, covering concepts such as that of injecting metal into the interstices around
space holder particles, which are subsequently removed, and the dispersion of
particles, which will release gas via chemical reaction or thermal decomposition.
However, in general, research on the production and performance of cellular metals remained at a low level until the 1990s, when a substantial acceleration took
place. It is now a research topic receiving a high level of attention and various industrial applications are currently being explored. Updated information about
these activities is available from certain websites [2] and a comprehensive review
has recently been published [3].
Further research is certainly necessary into the development of improved processing methods, since much of the material produced hitherto has been of relatively poor quality and/or inherently rather expensive. However, in this context it
is very important to understand the processing-microstructure-property inter-relationships for cellular metals and the relevance of these to the property combinations required for various applications. Making a foam from a metal, as opposed
to a polymer, boosts the stiffness, range of operating temperature and resistance
to many (organic) solvents, while, in comparison with a ceramic foam, important
advantages are expected in terms of toughness, (thermal and electrical) conductivity and formability. However, when considering the detailed characteristics, a clear
distinction should be drawn between open-cellular and closed-cellular metal, since,
not only are these two materials made by different processing routes, but in general rather separate types of application can be identified for them. Of course, all
types of cellular materials tend to be relatively light, and to have a high specific
stiffness, but these features usually depend primarily on the pore content, whereas
many other properties are much more sensitive to microstructure (cell structure
and nature of cell walls).
VII
VIII
Foreword
Closed-Cell Foams
Since the most promising methods for cheap production of bulk material tend to
generate closed-cell foams (commonly of aluminum), attention has been concentrated on these for many structural applications. These include various components designed to absorb energy progressively under relatively low applied (compressive) loads. In principle, the reduced constraint on cell walls, compared with
solid metal, should mean that large plastic strains can arise throughout via concertina-like deformation, with substantial absorption of energy. However, the performance of such components has often been a little disappointing hitherto, with premature failure commonly occurring within shear bands and such poor tensile ductility exhibited that components tend to fracture readily if any tensile stresses arise,
for example under bending moments. It is now becoming clear that these problems are substantially reduced if the cell size can be kept fine and uniform preferably to sub-mm levels. There is thus a strong incentive to develop processing
techniques capable of producing such material in bulk. Furthermore, it is commonly the case that the cell walls in these foams contain severely embrittling constituents, such as large ceramic particles and thick oxide films. Some such constituents are often deliberately introduced during processing in order to raise the
melt viscosity and thus inhibit cell coarsening and drainage [4]. Recent work has
confirmed that such constituents can have highly deleterious effects on the mechanical characteristics of the foam and has outlined the mechanisms responsible
for this [5]. Further work is required both on understanding these effects in more
detail and on developing processing routes in which these constituents are suitably
modified or eliminated. Closed-cell foams are also of interest for other types of application, such as thermal barriers, although ceramic materials would often be preferred for these.
Open-Cellular Metals
Suggested and actual applications include filters, catalyst supports, heat exchangers, fluid flow damping conduits (including various types of shock wave dissipation
devices), biomedical prostheses, internally-cooled shape memory actuators, air batteries, and protective permeable membranes and sheathes. Such functional components tend to incorporate higher added value than those in purely structural applications, which is appropriate in view of the generally higher costs of producing
open-cellular metal. The scale of the cell structure is often important for the functional characteristics. This will clearly be a basic specification for filters and fluid
flow limiting devices, but a fine cell size would often be preferred for heat exchangers etc. (subject to limitations imposed by any danger of pore clogging), while bone
in-growth into prosthetic implants might require relatively coarse pores. However,
these properties would often be needed in combination with a minimum strength
and ductility requirement, so a relatively fine, uniform cell structure and a defectfree cell struts' microstructure might be beneficial from that point of view. A wide
range of metals is being investigated for applications of open-cellular metals, bringing a requirement for improved understanding of process optimization issues in
various alloy systems.
Foreword
References
IX
Contents
1
2
2.1
2.1.1
2.1.2
2.1.2.1
2.1.2.2
2.2
2.2.1
2.2.2
2.2.2.1
2.2.2.2
2.2.2.3
2.2.2.4
2.2.2.5
2.2.3
2.3
2.3.1
2.3.2
2.3.3
2.3.4
2.3.4.1
2.3.4.2
2.4
2.4.1
2.4.2
2.4.3
2.5
2.5.1
2.5.2
XI
XII
Contents
2.6
2.6.1
2.6.2
2.6.2.1
2.6.2.2
2.6.2.3
2.6.3
2.6.3.1
2.6.3.2
2.6.3.3
2.6.3.4
2.6.4
2.7
2.7.1
2.7.1.1
2.7.1.2
2.7.2
2.7.2.1
2.7.2.2
2.7.2.3
2.7.3
2.7.3.1
2.7.3.2
2.7.3.3
2.7.3.4
2.7.4
2.7.4.1
2.7.4.2
3
3.1
3.1.1
3.1.1.1
3.1.1.2
3.1.1.3
3.1.1.4
3.1.2
3.1.3
3.1.3.1
3.1.3.2
Contents
3.1.3.3
3.2
3.2.1
3.2.2
3.2.2.1
3.2.2.2
3.2.2.3
3.2.2.4
3.2.3
3.2.4
3.2.4.1
3.2.4.2
3.2.5
3.2.5.1
3.2.5.2
3.2.6
3.3
3.3.1
3.3.1.1
3.3.1.2
3.3.1.3
3.3.1.4
3.3.2
3.3.2.1
3.3.2.2
3.4
3.4.1
3.4.2
3.4.2.1
3.4.2.2
3.4.3
3.4.3.1
3.4.3.2
3.4.3.3
3.4.3.4
3.4.3.5
3.4.3.6
3.4.3.7
3.4.3.8
3.4.4
3.4.4.1
3.4.4.2
XIII
XIV
Contents
4
4.1
4.1.1
4.1.1.1
4.1.1.2
4.1.2
4.1.2.1
4.1.2.2
4.1.2.3
4.1.2.4
4.1.2.5
4.1.2.6
4.1.3
4.1.4
4.2
4.2.1
4.2.1.1
4.2.1.2
4.2.2
4.2.2.1
4.2.2.2
4.2.2.3
4.2.2.4
4.2.3
4.2.3.1
4.2.3.2
4.2.4
4.2.4.1
4.2.4.2
4.2.5
4.3
4.3.1
4.3.2
4.3.3
4.3.3.1
4.3.3.2
4.3.3.3
4.3.3.4
4.3.4
4.3.4.1
4.3.4.2
4.3.4.3
4.3.5
Contents
4.3.5.1
4.3.5.2
4.3.6
4.3.7
4.3.7.1
4.3.7.2
4.3.7.3
4.3.8
5
5.1
5.1.1
5.1.1.1
5.1.1.2
5.1.1.3
5.1.2
5.1.3
5.1.3.1
5.1.3.2
5.1.4
5.1.4.1
5.1.4.2
5.1.4.3
5.1.5
5.1.6
5.1.6.1
5.1.6.2
5.1.6.3
5.2
5.2.1
5.2.2
5.2.2.1
5.2.2.2
5.2.3
5.2.4
5.3
5.3.1
5.3.2
5.3.3
5.3.3.1
5.3.3.2
5.3.3.3
XV
XVI
Contents
6
6.1
6.1.1
6.1.2
6.1.2.1
6.1.2.2
6.1.2.3
6.1.3
6.1.3.1
6.1.3.2
6.1.3.3
6.1.3.4
6.1.3.5
6.1.3.6
6.1.3.7
6.1.3.8
6.1.4
6.1.5
6.1.5.1
6.1.5.2
6.1.6
6.1.7
6.2
6.2.1
6.2.2
6.2.2.1
6.2.2.2
6.2.3
6.2.3.1
6.2.3.2
6.2.4
6.2.5
7
7.1
7.1.1
7.1.2
7.1.3
7.1.4
7.1.5
7.1.5.1
7.1.5.2
7.1.5.3
7.1.5.4
Contents
7.1.5.5
7.1.6
7.1.6.1
7.1.6.2
7.1.6.3
7.1.6.4
7.1.7
7.2
7.2.1
7.2.2
7.2.3
7.2.4
7.2.5
7.2.6
7.2.7
7.2.8
7.2.9
7.2.10
7.2.11
7.2.12
7.2.13
7.2.14
7.3
7.3.1
7.3.1.1
7.3.1.2
7.3.2
7.3.2.1
7.3.2.2
7.3.2.3
7.3.3
7.4
7.4.1
7.4.2
7.4.2.1
7.4.2.2
7.4.2.3
7.4.2.4
7.4.3
7.4.3.1
7.4.3.2
7.4.3.3
7.4.3.4
XVII
XVIII
Contents
7.5.4
7.5.4.1
7.5.4.2
7.5.5
7.5.5.1
7.5.5.2
7.5.6
8
8.1
8.2
8.3
Processing 355
Properties 358
Design and Application 360
7.5
7.5.1
7.5.2
7.5.3
7.5.3.1
7.5.3.2
7.5.3.3
Index 365
List of Contributors
O. Andersen
Frauenhofer-Institut fur
Fertigungstechnik und
Materialforschung (IFAM)
Auenstelle fur Pulvermetallurgie
und Verbundstoffe
Winterberstr. 28
01277 Dresden
Germany
M. Arnold
Lehrstuhl fur Werkstoffwissenschaften
Universitat Erlangen-Nurnberg
Martensstr. 5
91058 Erlangen
Germany
M. F. Ashby
Engineering Design Centre
University of Cambridge
Engineering Department,
Trumpington Street
Cambridge CB2 1PZ
UK
J. Banhart
Hahn-Meitner-Institute
Dept. of Materials SF3
Glienicker Str. 100
14109 Berlin
Germany
F. Baumgartner
Schunk Sinter Metalltechnik GmbH
Postfach 10 09 51
35339 Gieen
Germany
C. Beichelt
Wilhelm KARMANN GmbH
Karmannstr. 1
Postfach 26 09
49084 Osnabruck
Germany
T. Bernard
Neue Materialien Bayreuth GmbH
Universitatsstr. 30
94447 Bayreuth
Germany
H. W. Bergmann
Universitat Bayreuth
Lehrstuhl Metallische Werkstoffe
Ludwig-Thoma-Str. 36b
94440 Bayreuth
Germany
H. J. Bohm
Institute of Lightweight Structures and
Aerospace Engineering
Vienna University of Technology
Guhausstr. 2729
1040 Vienna
Austria
XIX
XX
List of Contributors
R. Braune
Lehrstuhl fur Fertigungstechnologie
Universitat Erlangen-Nurnberg
Egerlandstr. 11
91058 Erlangen
Germany
C. Haberling
Firma Audi AG
Abteilung
Werkstoffe/Verfahren/Recycling I/EG-34
85045 Ingolstadt
Germany
T. W. Clyne
Department of Materials Science
University of Cambridge
Pembroke Street
Cambridge CB2 3QZ
UK
F. Heinrich
Lehrstuhl fur Werkstoffwissenschaften
Universitat Erlangen-Nurnberg
Martensstr. 5
91058 Erlangen
Germany
T. Daxner
Institute of Lightweight Structures
and Aerospace Engineering
Vienna University of Technology
Guhausstr. 2729
A-1040 Vienna
Austria
Th. Hipke
Frauenhofer Institut fur Werkzeugmaschinen und Umformtechnik (IWU)
Reichenhainerstr. 88
09126 Chemnitz
Germany
H. P. Degischer
Institute of Materials Science
and Testing
Vienna University of Technology
Karlsplatz 13
1040 Vienna
Austria
B. Foroughi
Institute of Materials Science
and Testing
Vienna University of Technology
Karlsplatz 13
1040 Vienna
Austria
M. C. Hahn
Lehrstuhl fur Fertigungstechnologie
Universitat Erlangen-Nurnberg
Egerlandstr. 11
91058 Erlangen
Germany
C. Korner
Lehrstuhl Werkstoffkunde und
Technologie der Metalle
Universitat Erlangen-Nurnberg
Martensstr. 5
91058 Erlangen
Germany
J. Kovacik
Institute of Materials and Machine
Mechanics
Slovak Academy of Science
Racianska 75
P.O. Box 95
83008 Bratislava
Slovakia
A. Kottar
Institute of Materials Science
and Testing
Vienna University of Technology
Karlsplatz 13
1040 Vienna
Austria
List of Contributors
B. Kriszt
Institute of Materials Science
and Testing
Vienna University of Technology
Karlsplatz 13
1040 Vienna
Austria
R. Kretz
ARC Leichtmetallkompetenzzentrum
Ranshofen GmbH
Postfach 26
5282 Ranshofen
Austria
E. M. A. Maine
Centre for Technology Management
Engineering Design Centre
Engineering Department
University of Cambridge
Trumpington Street
Cambridge CB2 1PZ
UK
E. Maire
CR1 CNRS
GEMPPM
Batiment Saint Exupery
23 Avenue Capelle
69621 Villeurbanne cedex
France
U. Martin
Institute fur Metallkunde
TU Bergakademie Freiberg
Gustuv-Zeuner-Str. 5
09596 Freiberg
Germany
H. Mayer
Institute of Meteorology and Physics
University of Agricultural Sciences
Turkenschanzstr. 18
1180 Vienna
Austria
A. Mortensen
Laboratoire de Metallurgie Mecanique
Ecole Polytechnique Federale
de Lausanne
1015 Lausanne
Switzerland
U. Mosler
Institut fur Metallkunde
TU Bergakademie Freiberg
Gustuv-Zeuner-Str. 5
09596 Freiberg
Germany
C. Motz
Erich-Schmid-Institute of Material
Science
Austrian Academy of Science
Jahnstr. 12
8700 Leoben
Austria
R. Neugebauer
Frauenhofer Institut fur Werkzeugmaschinen und Umformtechnik (IWU)
Reichenhainerstr. 88
09126 Chemnitz
Germany
A. Otto
Lehrstuhl fur Fertigungstechnologie
Universitat Erlangen-Nurnberg
Egerlandstr. 11
91058 Erlangen
Germany
R. Pippan
Erich-Schmid-Institute of Material
Science
Austrian Academy of Science
Jahnstr. 12
8700 Leoben
Austria
XXI
XXII
List of Contributors
G. Rausch
Frauenhofer Institut fur angewandte
Materialforschung (IFAM)
Lesumer Heerstr. 36
28717 Bremen
Germany
R. F. Singer
Institut fur Werkstoffwissenschaften
Univertitat Erlangen-Nurnberg
Martensstr. 5
91058 Erlangen
Germany
C. San Marchi
Northwestern University
Department of Materials Science
and Engineering
Evanston IL 60202-3108
USA
S. Stanzl-Tschegg
Institute of Meteorology and Physics
University of Agricultural Sciences
Turkenschanzstr. 18
1180 Vienna
Austria
F. G. Rammerstorfer
Institute of Lightweight Structures
and Aerospace Engineering
Vienna University of Technology
Guhausstr. 2729
1040 Vienna
Austria
G. Stephani
Frauenhofer-Institut fur
Fertigungstechnik und
Materialforschung (IFAM)
Auenstelle fur Pulvermetallurgie
und Verbundstoffe
Winterberstr. 28
01277 Dresden
Germany
W. Seeliger
Wilhelm KARMANN GmbH
Karmannstr. 1
Postfach 26 09
49084 Osnabruck
Germany
M. Seitzberger
Institute of Lightweight Structures
and Aerospace Engineering
Vienna University of Technology
Guhausstr. 2729
1040 Vienna
Austria
F. Simancik
Institute of Materials and Machine
Mechanics
Slovak Academy of Science
Racianska 75
P.O. Box 95
83008 Bratislava
Slovakia
M. Thies
Lehrstuhl fur Werkstoffwissenschaften
Univertitat Erlangen-Nurnberg
Martensstr. 5
91058 Erlangen
Germany
B. Zettl
Institute of Meteorology and Physics
University of Agricultural Sciences
Turkenschanzstr. 18
1180 Vienna
Austria
List of Abbreviations
AFS
AVS
BET
CCD
CIP
COD
CT
CTOD
CVD
DOF
2D or 3D
EB-DVD
EDM
ESRF
FE
FORMGRIP
FPZ
HIP
IFAM
IMM
INSA
IP
IWU
LBM
LDC
LEFM
LKR
MIG
MMC
MURI
PCF
PM
PSF
XXIII
XXIV
List of Abbreviations
PVD
RV
RVE
SAS
SEM
TIG
US
XCT
1
Introduction: The Strange World of Cellular Metals
F. Simancik
When Nature builds large load-bearing structures,
She generally uses cellular materials: wood, bone, coral.
There must be good reasons for it.
M. F. Ashby
It is well known that porous structures are good for insulation, packaging, or filtering, but few people believe that they can also be very effective in structural applications. Thousands of scientific publications deal with the minimization of porosity
in load-bearing parts. Engineers work hard to eliminate pores from castings, powder metallurgy parts, weld joints, or coatings, thinking that a defect-free part is a
pore-free one. With this attitude it is difficult for someone to accept that a loadbearing material can include pores, even quite large ones. However, large natural
structures of porous materials have existed for thousands of years, demonstrating
how evolution has generated cellular structures that optimize mechanical properties and structural function for minimum weight.
Mankind tries to learn from nature. Understanding the benefits of natural structures gives us information to help us produce man-made cellular solids. The cellwall material has to be chosen very carefully if the structure is expected to carry
loads. Polymers appear to be insufficiently rigid and ceramics are too brittle. Perhaps metals could be the right choice. Several of the engineering properties of metallic foams are superior to those of polymeric ones: they are stiffer by an order of
magnitude, they are stable at elevated temperatures, they possess superior fire resistance, and they do not produce toxic fumes in a fire. Moreover, these materials
are fully recyclable without any pollution or waste problems. The latter fact can no
longer be ignored, because the production, disposal, and use of stronger and stiffer
materials in new products often have negative environmental impacts over the
product life cycle.
Owing to their pores, cellular metals possess a set of unusual properties
compared with bulk structural materials: they are crushable, they exhibit a plateau
stress if compressed, and they exhibit a change in Poisson ratio on deformation.
The excellent combination of good mechanical properties (mostly strength and
stiffness) and low weight is the prime advantage. In addition, cellular metals
absorb high impact energies regardless of the impact direction, and are very
efficient in sound absorption, electromagnetic shielding, and vibration damping.
Most of the mechanical properties of foam materials can be achieved with other
materials, sometimes more effectively, but foams can offer a unique combination
of several (apparently contradictory) properties that cannot be obtained in one conventional material at the same time (e. g., ultra-low density, high stiffness, the capability to absorb crash energy, low thermal conductivity, low magnetic permeability,
and good vibration damping). Cellular metals are thus promising in applications
where several of these functions can be combined.
These properties depend significantly on the porosity, so that a desired portfolio
of properties can be tailored by changing the foam density. This is one of the most
attractive features of these remarkable materials.
Cellular material properties also depend on the pore structure. This influence,
imperfectly understood at present, is a topic of intense study. Various constitutive
laws have been suggested for the characterization and modeling of this relationship. These laws, originally developed for polymeric foams, are usually based
on the relative density of the foam, and therefore suppose uniform cellular structure, at least at a macroscopic level. However, metallic foams are dramatically different from polymeric foams: polymeric foams generally have a regular microstructure, whereas metallic foams may be highly disordered with a wide dispersion of cell size and shape. Moreover, many imperfections exist in a cell
structure, such as cracks or holes in the cell walls, corrugated cells etc. These
effects are inevitable due to manufacturing at significantly higher temperatures
than in the case of polymers. If these features are not taken into account and
the properties of the foam are characterized only in relation to apparent density,
a higher scatter of properties is to be expected. This is why it is still widely
believed that acceptable reproducibility of the properties of metallic foams is
questionable.
The structure of metallic foams is often non-uniform, especially in the case of
complex 3D parts. It should be noted that a uniform structure is not necessary
for obtaining acceptable and reproducible properties. Anisotropic or gradient
pore structures allow the distribution of load bearing material according to load
conditions (simulating the optimum bone-like structure), without a need to increase the overall weight or volume of the component. Therefore, the challenge
for manufacturing is not to produce a uniform structure, but to achieve reproducible properties with a controlled non-uniform structure.
If the non-uniform structure is optimal, the crucial question What is the material? should be answered. It is really difficult to distinguish between material and
structure. If a cellular metal is a material, it is very problematic to define geometryindependent material characteristics (the strength or elasticity modulus); if it is a
structure made of a certain metal it is almost impossible to define its random
geometry.
Cellular metals can be prepared by various processing methods. They may all be
called metallic foams, but they are very different materials, depending on the
manufacturing technique. The production method affects the distribution of the
cell-wall material in such a way that the properties of differently manufactured materials are not comparable.
Metallic foams result from the nucleation and subsequent growth of gas bubbles
in a liquid or semi-liquid metal. They usually have a non-uniform pore structure
(variable pore size and sometimes preferred orientation of pores). The pores are
initially closed, but some defects always appear on cooling, owing to shrinkage of
solidifying metal and gas-pressure reduction in pores. Many solidified cell faces
have non-uniform curvature or are corrugated and have occasional broken walls
that still hang in place. These are the main features of this kind of cellular solid.
Other cellular metals may be manufactured by casting or deposition of the metal
onto templates or place holders, which have to be removed from the final product,
thus creating a porous structure. These cellular structures usually have adjustable
distribution of pore size (according to the template) and their cells are always open.
The manufacturing process dictates not only the properties but also the potential
applications of the foam. Thus foams prepared by the powder compact foaming
(PCF) technique (usually with a dense skin) can be effectively used as net-shape
components, stiffening cores in castings, or in complicated hollow profiles,
whereas the foams prepared by the molten metal route (typically large blocks
or panels) can be effectively used as voluminous energy absorbers, cores for sandwiches, or for blast protection. The open cellular structures made by investment
casting are good for heat exchangers, sound absorbers, or for electrodes in batteries. The properties arising from the cellular structure produced by a certain
manufacturing technique cannot be effectively achieved using another method.
This also means that cellular metals manufactured differently are not necessarily
competitive materials.
The first attempt to foam a metal was performed by B. Sosnik in 1943 [1]. In
order to create pores, he added mercury to molten aluminum. In 1956 J. C. Elliot
replaced mercury by foaming agents generating gas by thermal decomposition [2],
so now modern scientists and engineers can develop metallic foams without
having to deal with the toxicity of mercury.
In 1959 B. C. Allen [3] invented the PCF route for manufacturing metallic foams
and the basic processing techniques were thus completed.
The success in the preparation of the first metallic foams and discovery of their
remarkable properties started a euphoric enthusiasm for these materials. In 1957
J. Bjorksten stated [4]: Foamed metals offer great market potential and might conceivably account for 10 % of all metals produced within 20 years. However he also
said: A lot of work still remains to bring about production on a large scale, such as
closer control of density and dimensions. Unfortunately, only the second of his
statements turned out to be realistic.
Although it is many years since the first patents concerning the manufacture of
metallic foams appeared, the material has not been put into large-scale commercial
production yet. This discouraging fact can be attributed to inadequate design of
components, low reproducibility of properties, a lack of testing procedures and calculation approaches, absence of concepts for secondary treatment, as well as the
production technologies being too complicated and relatively expensive.
References
2
Material Definitions, Processing, and Recycling
H. P. Degischer
Processing techniques for cellular metals classified according to the state of the
metal, the formation of the cellular architecture, and the pore-forming ingredients.
Figure 2-1.
density of less than 0.3 [1]. Materials with higher relative densities are called porous materials (for instance, powder compact greens), most of which also can be
produced by the processing techniques listed in Fig. 2-1. The most widely developed and investigated cellular metals are based on aluminum and its alloys.
The general aspects of processing are described elsewhere [1 3] and specific presentations can be found in the proceedings of the conferences dealing with cellular
metals [4 10]. The most important processing techniques are described in detail in
subsequent chapters.
Cellular metals are heterogeneous materials formed by a 3D metallic matrix with
gas-containing pores occupying more than 70 vol.-%. Cellular metals are classified
according to the following criteria (Fig. 2-1).
x
Open porosity structures can be formed by replication (Duocel [6]), deposition (Incofoam [6]) or by construction of solid ingredients with space between (for instance, the 3D networks of lattice block materials (LBM) [5,6] or those prepared
by rapid prototyping techniques [11]). Open-cell solid structures may be called
sponges. Closed cells are produced by embedding or cementation of hollow ingredients (syntactic foams), or by foaming in the liquid state.
The expression metal foams, strictly valid only for the liquid phase, is often
used to describe the solid product. Mixtures of metal powders and blowing agents
are compacted by extrusion or hot pressing providing a precursor material foamed
above the solidus temperature, a method called powder compact foaming (Alulight [5 10], IFAM-Foaminal [3,5 10], Alufoam [5,7]). The Formgrip material
[3,7] is made by remelting a stir-cast foamable precursor metal matrix composite.
Low-density-core material (LDC [3,6]) is produced by pore formation in the solid
state by the high gas pressure of entrapped dissociating blowing agents. A blowing
agent powder is mixed into the melt either in a crucible (Alporas [17]) or in the
gate in pressure die casting (Buehler [10]). Foaming of particle-reinforced metals
takes advantage of the stabilization of gas bubbles by the ceramic ingredients producing a cellular metal matrix composite from the melt (Cymat [1 7], COMBAL
[10]). The formation of a gas metal eutectic is the principle for the production
of Gasar foams [1,2,6]. A slip reaction foam technique based on foaming of precursor slurry by chemical reaction and a reaction sintering process for aluminides has
been described [10]. The originally closed cells of foamed metals may not be gas
tight after solidification owing to cracks in their walls.
The resulting cellular metal products can be differentiated by their structural features. The term structure is used for the description of cellular materials at different levels of observation (structology): the geometric architecture of the solid
(skeleton) in the individual cells and their 3D arrangement, the variations of that
References
2.1
The melt route for processing closed-cell metal foams is very attractive since this
approach allows economic handling of large quantities of material. Melt-route processes are also well suited to the use of scrap as feedstock. In order to foam the
melt properly, gas must be introduced. This can be either done by gas injection
or by in-situ gas generation by a chemical decomposition of a foaming agent.
For the production of homogenous foams some prerequisites have to be fulfilled.
If foaming is by in-situ gas generation uniform dispersion of the foaming agent in
the melt within a time that is short compared to the decomposition or reaction
time of the additive is required. In addition, the escape of gas during the foaming
process has to be prevented. One important step to meet these requirements is to
increase the viscosity of the melt. There are several approaches to do this: foaming
in the semi-liquid state, incorporation of ceramic particles [1] or oxidation [2]. The
effect of the particles added is always twofold: beside increasing the melt viscosity
they stabilize the cell walls.
Independent of the specific foaming process it is found that porosity, quantified
by density, and mean cell size are intimately related. Generally, for a particular alloy
or composite the mean cell size can not be chosen independently of the porosity.
2.1.1
The Cymat/Alcan and Norsk Hydro melt-foaming process is a continuous, gas-injection method developed simultaneously and independently by Alcan [1] and
Norsk Hydro [3] in the late 1980s and 1990s. A sketch of the process developed
by Alcan is given in Fig. 2.1-1. The patent is now licensed and exploited by the
Cymat Aluminum Corporation [www.cymat.com].
The process employed by Hydro Aluminum, Norway is analogous. A metal matrix composite (Al-wrought or Al-casting alloy matrix 10 30 vol.-% SiC or Al2O3
Figure 2.1-1. Principle of the melt-foaming route employed by Cymat. The foam casting process
for producing flat panels consists of melting and holding furnaces, the foaming box and foaming
equipment, and a twin-belt caster [1].
particles) is used as a starting material. The starting material is molten with conventional foundry equipment and transferred to a tundish where gas, typically air,
is injected via small nozzles incorporated into a rotating impeller, thus forming a
dispersion of small gas bubbles. The bubble size can be controlled by adjusting the
gas flow rate, the impeller design (number of nozzles and their size), and the speed
of rotation of the impeller. The gas bubbles rise to the surface where they accumulate. The ceramic particles are trapping gas bubbles owing to the favorable interface
energy and serve as stabilizer of the cell walls and delay their coalescence. They
also reduce the velocity of the rising bubbles by increasing the viscosity of the
melt. That is, the particles reduce the kinetic energy of the rising bubbles and
hence the danger of mechanical rupture when they arrive at the surface. The resulting metal foam, which is still liquid, is carried away by means of a conveyor
belt where it solidifies and cools. The relative density is predominantly controlled
by the process parameters, such as rotor speed, gas flow, and the amount of particles in the melt, and finally the solidification condition.
This process of casting aluminum foam is capable of producing slabs with a relative density in the range 2 20 % (0.05 0.55 g/cm3). The average cell size is inversely related to the density (Fig. 2.1-2a) and is in the range 2.5 30 mm [1].
a)
Figure 2.1-2.
b)
10
Figure 2.1-3. Optical micrographs of Cymat foam produced by the gas-injection method: left)
cell structure (density about 0.3 g/cm3), inhomogeneous and anisotropic; right) foam surface
(density about 0.05 g/cm3).
The production facility set up by Cymat is capable of casting foam panels in continuous length at an average rate of 900 kg/h up to 1.5 m wide with a thickness
range of 25 150 mm. This shows that the process is relatively straightforward
and economical. Challenges that may require more work in the future include
the variation in cell size, the density gradient, and the anisotropy of the cell structure, which results from mechanical forces from the conveyor belt (Fig. 2.1-3).
Especially earlier prototype have been compressed by the conveyor belt producing
flattened pores causing poor stiffness and strength values along the thickness of
the slabs. It should also be noted that the preparation of the composite feedstock
requires relatively tedious long time stirring processes to achieve the proper homogenous distribution of the particles in the melt.
In principle, this foam generation technology also allows the casting of nonrectangular, 2D profiles as well as 3D shapes.
2.1.2
In-situ Gas Generation: the Shinko Wire Process and the FORMGRIP process
The gas-injection method suffers from the fact that a relatively small number of
large bubbles is generated, which leads to rather coarse and irregularly shaped
pore distribution. Two techniques are described below, in which the foaming gas
results from a thermal decomposition of solid ingredient. In this way a huge number of bubble nuclei is created throughout the melt.
Figure 2.1-4.
the viscosity by the formation of oxides: CaO, Al2O3, CaAl2O4. There is an appropriate stirring resistance for optimizing the foaming ratio [4]. The thickened aluminum is poured into a casting mold and stirred with an admixture of 1.6 % TiH2 as a
foaming agent. While vigorously stirring, the TiH2 dissociates and H2 -bubbles are
formed causing molten material to expand and to fill the mold. Then, the foamed
material is cooled by fans to solidify in the casting mold. After removal from the
casting mold. An Alporas block 450 mm wide, 2050 mm long, and 650 mm
high is sliced into plates.
Alporas is an ultra-light material with a closed-cell architecture (Fig. 2.1-5). The
density of the product is 0.18 0.24 g/cm3, the mean cell size is about 4.5 mm. Alporas is regarded as the best commercially available aluminum foam in terms of
regular cell micro structure. This is to a certain extent because transfer and deformation of the still liquid foam, as in the Cymat/Norsk Hydro process, is avoided.
The cell architecture is the outcome of a growing process where bubble expand and
coalesce due to cell wall rupture.
Figure 2.1-5. Typical cell structure of an Alporas foam. As a result of the growth process most
cells are far from equiaxed. The global homogeneity is superior to other commercially available
aluminum foams.
11
12
Figure 2.1-6. Diagram of the melt-based FORMGRIP process for production of near net-shape
metal foam parts [7].
Figure 2.1-7.
Figure 2.1-8. Examples of cross sections of aluminum alloy FORMGRIP foams baked under
different conditions. Porosity, P, levels and the mean cell sizes, d, are: a) P 69 %, d 1.1 mm;
b) P 79 %, d 1.9 mm; c) P 88 %, d 3.1 mm [7].
The relation between relative foam density and average cell diameter fits that derived for Cymat foams. Fig. 2.1-2b shows the mean cell diameter as a function of
the reciprocal density (see Section 4.1). The mean cell diameter is inversely proportional to the density indicating that foam expansion is governed by cell coalescence
and the mean cell wall thickness is constant. The dependence of the cell diameter
on the density is the same for both contents of SiC particles. The influence of particle size has not yet been investigated. Theoretical work of Kaptay [9] and experimental work of Weigand [10] indicate that a reduction of particle size will not lead
to a higher stabilization and therefore to a smaller critical cell wall thickness.
In terms of geometrical complexity and simultaneous microstructural control,
the FORMGRIP process surpasses the Shinko Wire one. There is no need to transfer material from a mold into a die while the foaming process is going on. In addition, the precursor material can be shaped before the final baking step. The economics of the FORMGRIP process, however, are clearly inferior to the other processes discussed in this chapter. This is due to the discontinuous nature of the process as well as its various number of processing steps.
13
14
References
2.2
There are many different ways to manufacture cellular materials [1]. One of the
available processes has become increasingly popular in the past few years and is
at the stage of industrial implementation now. The method is sometimes loosely
called the powder-metallurgical route, but the term powder-compact foaming
technique seems more appropriate.
2.2.1
The technique consists of mixing aluminum or aluminum alloy powders with appropriate foaming agents, which get entrapped by compacting this mix to a dense
product called foamable precursor material. The powder mix can be compacted directly by hot pressing, conform extrusion, or powder rolling. Alternatively the pow-
Figure 2.2-1.
der may be cold compacted for better handing in conventional extrusion or rolling
(Fig. 2.2-1). Heating the precursor above its solidus temperature releases the pressure on the foaming agent allowing decomposition and formation of bubbles. After
cooling a low-density foam structure of originally closed cells is obtained [2,3].
The method is not restricted to aluminum and its alloys: tin, zinc, lead, gold, and
some other metals and alloys can also be foamed by choosing appropriate foaming
agents and process parameters (see Section 2.1.2). The most common alloys for
foaming, however, are pure aluminum or wrought alloys such as aluminum
2xxx, 6xxx, or 7xxx alloys, e. g. AA 2014, 6060, 6061, 6082, or 7075. Casting
alloys such as AlSi7Mg (A356) and AlSi12 are also frequently used because of
their low melting point and good foaming properties, although in principle virtually any aluminum alloy can be foamed by carefully adjusting the process parameters.
Quite complex-shaped metal foam parts can be manufactured by expanding the
foam inside a mold, thus confining spatial expansion. An example for one such
part is shown in Fig. 2.2-2. The part, developed in the framework of a feasibility
study, is a novel pantograph horn for an electrical locomotive. This light-weight solution based on aluminum foam replaces traditional cast aluminum parts saving
30 % weight.
A nice feature of the technique is that composite structures consisting of an aluminum foam and bulk metal parts can be made without using adhesives. Examples are foam-filled aluminum sections and sandwich panels with an aluminumfoam core and metallically bonded steel, aluminum, or even titanium face sheets.
For making such composites the foamable precursor material is first bonded to the
solid section or sheet by co-extrusion or roll-cladding, after which the foamable
core layer is expanded by heat treatment [4,5] (see also Section 3.3).
15
16
The advantages of the powder-compact route are obvious and are listed in
Table 2.2-1. Beside the first two features already mentioned, the flexibility arising
from the preparation of the precursor from powders is important. Alloys can be
made simply by mixing inexpensive elementary powders. No ceramic additives
are needed to stabilize the foam, in contrast to some of the melt-route foaming processes in which up to 15 % silicon carbide has to be added [1,6]. However, if required, ceramic powders, metal fibers, or ceramic fibers can be added to the powder blend for special applications, such as for reinforcement or to increase wear
resistance. Naturally, there are also some disadvantages that are inherent to the
process. Metal powder is more expensive than bulk metal and requires effort for
compaction. This rules out applications that require very cheap materials. Moreover, the size of aluminum foam parts that can be manufactured is limited by
the size of the baking furnace, and is therefore smaller than for some of the competing melt-foaming processes. The largest sandwich components that have been
manufactured using the powder-compact foaming technique are about 2 m q 1 m
q 1 cm in size (possibly larger in future). At LKR in Ranshofen a part of similar
size was produced without face sheets (Fig. 2.2-3). True 3D-volume parts are
usually not thicker than 30 cm, a limit which is difficult to shift to higher values.
A large aluminum foam column produced at Fraunhofer IFAM was 1 m high and
18 cm in diameter, weighing 13 kg. In contrast, the liquid-metal route allows for
making panels 15 m in length [7] and 100 cm thickness [8]. However, as these processes cannot be used for near-net-shape production and only permit very simple
geometries, they are appropriate for different fields of application. Continuous
foaming of long products is under investigation [9].
The middle column of Table 2.2-1 lists some of the problems that are still encountered when foaming aluminum with the powder-compact melting method
but which can, in principle, be solved with further research.
Advantage
Problem
Disadvantage
Cost of powders
Composites can be
manufactured
Permeable (holes)
Coating process
requires sealing
Difficult to control
Figure 2.2-3.
2.2.2
17
18
be of sufficient quality. However, powders from different manufacturers led to notable differences in foaming behavior and empirical criteria have been derived to
facilitate the selection of powders. The cost of powders and the ability of a manufacturer to provide sufficient quantity with a constant quality are also crucial.
As already pointed out, alloys can be obtained in different ways. The frequently used alloy AlSi7, for example, can be either prepared by atomizing a
AlSi7 melt, or by blending pure aluminum powder with 7 wt.-% silicon powder,
or, in a third way, by mixing 58 % of standard AlSi12 powders with 42 % aluminum powder.
Mixing
The mixing procedure should yield a homogeneous distribution of alloying elements and the foaming agent to ensure that high-quality foams with uniform
pore-size distributions are obtained. Powders are mixed in batches of 500 kg at
Schunk-Honsel in commercial large-scale tumbling mixers with parameters determined in technological tests. Alternatively, powder mixes can be obtained by aerodynamic mixing. For example, Alulight International GmbH Austria mixes aluminum and titanium hydride in large containers with 50 80 short pulses of pressurized nitrogen gas.
2.2.2.2
Densification
Powder consolidation can be carried out by various techniques. It has to be ensured
that the foaming agent is completely embedded in the metal matrix and no residual open porosity remains. One way to obtain foamable precursor material with
nearly 100 % theoretical density is the combined use of cold isostatic pressing
(CIP) and ram extrusion. CIP is first applied to consolidate the powder mix to billets with a relative density of 70 80 % and a mass of typically 50 kg. These billets
are used in the subsequent extrusion step. Although CIPping is not absolutely necessary (powders have been filled into thin-walled aluminum cartouches and inserted into the extrusion machine without prior consolidation) it has additional advantages such as the prevention of powder contamination and powder de-mixing.
The CIP billets themselves are not foamable because of their large content of residual porosity, which causes massive hydrogen losses when the material is heated.
To obtain foamable material, the billets are preheated to 350 hC and extruded as
rods or any other profile. For this a horizontal direct extrusion machine is used
(25MN Schunk-Honsel). The extrusion machine is operated in cycles with a new
billet inserted after each extrusion step. This way rather high outputs can be
achieved.
Foamable material has also been manufactured by rotary continuous extrusion
in the so-called CONFORM process by Mepura (Ranshofen) [10]. Here a rotating
wheel is used to drag the powder into the consolidation chamber from which it is
pulled off in radial direction as a compacted wire. Foamable wires of about 8 mm
diameter were manufactured from wrought alloys containing titanium hydride.
2.2.2.3
Foaming
Heat treatment at temperatures above the solidus temperature of the foamable matrix is necessary to produce the foam structure. The gas released by the decomposing foaming agent may form pores in the solid state but only above the solidus are
bubbles formed and the matrix expands up to a maximum volume, that is to a
minimum density. The density and density distribution of the growing foam can
be controlled by several parameters. The foaming agent content in the precursor
material is obviously important, but furnace temperatures and heating rates also
have an influence [12]. The mold material, the mold shape, and the type of furnace
naturally influence the heating rate and have therefore also to be considered. A
careful control of the heating conditions during foaming is essential for obtaining
high-quality foams. The difficulty is that the liquid foam is thermodynamically unstable and conditions change constantly during foaming. There are various intermediate stages: at first only the mold is heated directly, whereas the foamable material receives heat only indirectly via heat conduction through the mold. Initially
there are merely some point contacts between the piece of foamable material in the
mold and the mold walls. However, as the temperature increases, the precursor
softens and assumes the contour of the mold thus increasing the transfer of
heat. Moreover, heat transfer via radiation gains importance with rising temperatures. The reflectivity of the mold and precursor surfaces may change during the
process and add a further variable. Finally, after foaming has started, the thermal
conductivity of the precursor rapidly decreases thus reducing heat flow. As soon as
the mold has been filled with foam it has to be cooled down below its solidus temperature to stabilize the foam structure. The phenomena during cooling are also
quite complex and difficult to describe for reasons similar to those mentioned
for the heating phase.
Typical densities of aluminum foams are in the range 0.4 0.8 g/cm3 including
the closed skin around the foam body. The final density of a foamed part can be
simply predicted if the volume of the hollow mold and the mass of the inserted
precursor material are known. The foaming mold may be loaded with several
small pieces or one single piece of precursor. Choosing the latter method (which
is preferred by LKR and SAS [13]) one has to take into account that each piece
of the expanding precursor material has a dense aluminum oxide layer on its sur2.2.2.5
19
20
Left) top of aluminum foam part made by inserting various pieces of foamable
material into the mold (dark figures indicate original size of the precursor pieces). Right) foam
part made of two pieces of precursor without achieving bonding between the two pieces.
Figure 2.2-4.
face, which has to be broken up by expansion of the individual foam pieces. Incomplete foaming may cause the foamed pieces to remain separated even after the
foaming process (Fig. 2.2-4b). A relative movement of the foam pieces to each
other helps to break up the oxide films. Fig. 2.2-4 shows an example of a successful
formation of a foamed body from various pieces of the precursor and an example
of failure. In the former case the location of the original individual foam pieces can
still be identified from the contrast in gray scales between the various regions:
darker gray identifies oxide layers of extruded surfaces, brighter gray is the new
(expanded) surface. This effect is currently exploited to create foam panels and
other foamed parts for making designer objects.
2.2.3
State of Commercialization
Currently the foaming technique described is still in the stage of industrial implementation. Nevertheless, a number of companies have already made commitments
for a future production and are building up facilities [14]. The joint effort of Schunk
Sintermetalltechnik (Gieen) and Honsel GmbH&Co KG (Meschede) is one example. Owing to their collaboration with Karmann the activities are preferentially directed towards foam and foam sandwich parts with a complex 3D geometry (see
Section 3.3). Alulight International GmbH is another example. It is a joint venture
of SHW (Germany) and Eckart Austria. The company offers aluminum foam
panels in sizes up to 625 mm q 625 mm, with thickness of 8 25 mm. Neuman
Alufoam, another Austrian company, also offers foamable precursor material
(extrusions) and foamed parts.
References
2.3
The previous section was dedicated exclusively to aluminum foams. For many applications one would like to use cellular materials made from metals or alloys other
than aluminum. There have been some attempts to manufacture metal foams by
simply adapting the powder-compact process originally developed for aluminum to
other metals by adjusting the properties of the foaming agent and the process parameters. This procedure was successful in some cases. However, for high-melting
alloys the powder-compact foaming technique is difficult to implement and especially for titanium no promising results could be obtained. Here alternative routes
based on advanced powder metallurgy yielded better results. Therefore, in the current section the topic will be slightly extended from foamed to cellular or porous
metals in a more general sense.
21
22
2.3.1
Zinc
Zinc can be foamed by a straight-forward modification of the powder-compact technique. The foaming agent used for aluminum (TiH2) can be used, although ZrH2
seems to yield slightly better results. Powder properties and mixing procedures are
quite similar to aluminum. Only the pressing and foaming temperature has to be
chosen slightly lower than for aluminum due to the melting temperature of zinc at
419 hC. Foamed zinc shows a very uniform pore structure. This can be attributed to
the fact that the decomposition temperature of the foaming agents TiH2 matches
with the melting temperature of the metal. Therefore, melting and pore formation
occur simultaneously and round bubbles are created from the very beginning. In
contrast to aluminum, there is no solid state expansion range with corresponding
crack formation. Fig. 2.3-1a shows an example of a zinc foam.
2.3.2
Lead
Lead and lead alloys such as Pb Sn and Pb Sb can be foamed by another modification of the process. TiH2 and ZrH2 cannot be used as foaming agents because
of the low melting temperatures of pure lead (327 hC) and even lower solidus temperature of the alloys. Quite good foams have been obtained by using lead(II) carbonate as a foaming agent: it decomposes above about 275 hC and releases CO2 and
water, which act as foaming gas. Fig. 2.3-1b shows an example of a lead foam
2.3.3
Titanium
Owing to its high melting temperature (1670 hC) and relatively low density (4.51 g/
cm3), titanium and its alloys are excellent materials for lightweight applications at
elevated temperatures and are widely used in aeronautical applications. Porous ti-
a)
Figure 2.3-1.
b)
tanium structures have an additional potential for weight reduction and could even
be suitable for functional applications if the pore structure were open.
In principle, there are many possible production methods for cellular materials
based on titanium (see Section 2.4), most of them starting from metal powders.
1.
2.
3.
4.
Figure 2.3-2. Open porous titanium made by space-holder technique: left) pore size 1 4 mm;
right) pore size about 500 mm, porosity 55 80 %.
23
24
Figure 2.3-3.
Figure 2.3-4. Strength and Young's modulus as a function of density obtained from bending and
tension tests [5].
of the space-holder powder, spherical and angular pores in the range 0.1 2.5 mm
can be obtained, resulting in overall porosities of 70 80 %. It was found that the
sintering activity can be increased by partially substituting titanium by titanium hydride, thus yielding an increased compression strength.
At Fraunhofer IFAM, polymer granules were used as the space holder. They
were removed by a chemical process at temperatures around 130 hC, after pressing.
After space-holder removal, samples are sintered in vacuum at temperatures of
1100 1250 hC. Depending on the particle size of the granules, average pore diameters in the range 200 3000 mm can be obtained. Fig. 2.3-2 shows some typical
samples. Fig. 2.3-3 shows the typical pore structure of samples based on spherical
space-holder granules. Beyond the primary pore structure, some microporosity
(secondary pores) inside the sintered network is visible. It was shown that the secondary porosity has a strong influence on the overall strength of the samples and
can be reduced by either changing the sintering parameters and/or partially replacing titanium powder by titanium hydride [4]. As for all porous materials, the mechanical properties of cellular titanium are a function of density. Fig. 2.3-4 shows
the strength and Young's modulus obtained from bending and tension tests as a
function of density.
2.3.4
Steel
Powder-Compact Foaming Technique
The long experience in making aluminum foams from powder metallurgy (PM)
precursors encouraged researchers to transfer this process to higher-melting
materials such as iron-based alloys and steels. The major requirements for an
adaptation of the foaming agent process to this group of materials are the following.
2.3.4.1
x
x
x
x
The basic requirements for foaming agents are: point of gas emission above
1000 1200 hC (depending on the alloy composition), broad temperature range of
gas emission (up to 1550 hC for nearly pure iron), and sufficient volume of gas release. It was found that especially metal nitrides and certain carbonates show a significant gas emission and qualify for being useful as foaming agents. Examples are
manganese nitride, chromium nitride, molybdenum nitride, calcium carbonate,
strontium carbonate, and barium carbonate [5,6]. Theoretical investigations [5,7]
have shown that both the iron carbon and the iron boron system [7] are able
to meet the basic requirements for being foamed to iron-based metallic foams,
namely: a low melting point matching the decomposition temperature of the foaming agent, and a broad two-phase semisolid region in the phase diagram, thus
creating a wide foaming interval. As for the production of Al based foamable pre-
25
26
Iron-based foams
obtained from the foaming agent
process.
Figure 2.3-5.
cursor material, extrusion has been successfully used for compacting iron powder
mixtures. The resulting samples are shown in Fig. 2.3-5.
Experiments with powder mixtures of iron and carbon have shown that free carbon without any additional foaming agent already leads to a certain degree of porosity. Carbon is oxidized during the foaming process and the resulting gaseous
CO and CO2 creates pores. However, pore size distributions are not uniform and
pore shape is usually rather irregular. The porosity mainly results from large, isolated pores (Fig. 2.3-6). Adding 0.25 % SrCO3 leads to an increase of porosity to
55.5 % (Fig. 2.3-7). The pore structure at this composition appears to be more
homogeneous and the average pore size is obviously lower. Increasing the amount
of SrCO3 results in a further increased porosity (64.3 %). From that it can be concluded that SrCO3 has a significant influence on the achievable porosity and the
maximum expansion. The foaming agent technique has therefore been shown to
be feasible for steel. However, foaming of stainless steel or even superalloys has
not yet been successful and the general state-of-the-art of foaming steel with the
foaming agent method is still far behind the aluminum foaming technology.
Figure 2.3-6.
Figure 2.3-7.
SEM images of sintered specimen: left) stainless steel 316L, 1100 hC, 1 h, particle
size I16 mm; right) Inconel 600, 1250 hC, 1 h, particle size 100 200 mm [4].
Figure 2.3-8.
27
28
References
2.4
H. P. Degischer
Cellular metals compete with polymers for some applications. The recyclability of
metals is one of their benefits, enabling ecologically sustainable product life cycles
[1]. Compared to bulk metal products, there are two complications to be tackled in
remelting cellular metals:
x
The high surface-to-volume ratio of the order of 100/length unit increases the extent of surface adsorptions and reactions.
The low average density, due to the high porosity filled with gas, makes the cellular material float on its melt.
2.4.1
Heating cellular aluminum in air enhances the growth of the oxide film not
only in open cellular structures, but also in those foamed by formation of
closed pores in the liquid, or semi-solid state, which usually become permeable
to gas after solidification and cooling. When heating the cellular aluminum up
to the melting point in a short time, about 1 h, the thickness of the oxide film
may reach about 10 mm and any cracks formed will be covered quickly by
ongoing oxidation. The growth will slow down after prolonged heating, when
the thickness reaches about 100 mm [2], which is similar to the cell-wall thickness.
The dehydrated oxide is stable up to about 2000 hC, so the cellular aluminum part is
slowly converted to a cellular alumina structure, which may maintain its macroscopic shape if it does not break up under its weight, or an external force.
It is difficult to submerge a cellular part in a melt because it tends to float
and contains a lot of air, which would have to be replaced by melt. Therefore
the cellular structures have to be compressed as much as possible, as is done
with packaging foils, or shredder scrap before melting [1,3]. Cellular metals
can be shredded too and treated as normal shredder scrap. Nevertheless, the
high content of surface oxides reduces the efficiency of metal reclamation.
The only experience is from small batch laboratory trials on scrap from powder-compact foamed aluminum [4], in which efficiencies up to 80 % have
been reached. Oxides and other impurities have to be removed by the usual
cleaning process for melts to the extent necessary for the later production process [3]. The purity requirements may be very low for the production of cellular
metals. Impurities in precursor material prepared for powder-compact foaming,
as well as for foaming in the melt, might be advantageous, as long as they act
as nuclei for the formation of pores. There might be even an up-grading of lowquality scrap when used for the production of foamed metals [5]. The Ti remaining from the blowing agent (usually I0.5 wt.-%) is soluble below 0.12 wt.-% at
the peritectic temperature of the binary aluminum melt [1] and does not degrade the quality of the alloy, but provides a grain-refining effect during solidification. ALPORAS foam contains about 2 wt.-% Ca, which is an element to be
restricted below 0.1 wt.-% in all cast alloys. Ca will be partly oxidized and thus
transferred into the dross, but some care has to be taken in the mix of ingredient scrap for not exceeding the specified impurity levels for the secondary
alloy. The impurity levels of secondary aluminum cast alloys are not as stringent
[1,3], so that scrap of the common types of cellular aluminum can be added to
the remelting furnace.
2.4.2
29
30
for the production of metallic foams, may not be as stringent as placed on MMC
for bulk components, thus higher quantities of oxide skins and other impurities
may be tolerated, or are even advantageous in foamable aluminum. Figure 2.4-1
shows the material flow cycle for discontinuously reinforced metals and cellular
metals made from MMC indicating the possible interactions between these two
cycles:
x
Remelted MMC, with, or without cleaning treatments, can be used as a base material for the melt foaming processes and for FORMGRIP.
Any secondary alloy, including that reclaimed from MMC melts, can be used for
production of cellular structures by any of the known production methods.
Scrap from MMC foams can be introduced into the MMC cycle, if quality
requirements are met.
The same rules have to be obeyed for the recycling of MMC foams as for the MMC
matrix [11] in addition to the pretreatment (compaction and drying) to reduce gas
evolution. Two main problems have to be tackled when remelting MMC.
x
x
The reactivity of the reinforcement with the melt increases with temperature.
The dewetting tendency of the reinforcement limits conventional melt cleaning
methods.
Figure 2.4 -1. Recycling of discontinuously reinforced metals and of metal foams based on MMC,
indicating the reuse of secondary MMC for the processing of foamed metals either by the powdercompact method, or by the melt-foaming technique.
The reaction between 6xxx type wrought alloys and alumina reinforcements is
driven by the Mg content forming MgAl2O4, spinel. An original level of approximately 2 vol.-% spinel seems to stabilize at just above 3 vol.-% after several remelting cycles; this does not influence the mechanical properties significantly [12], but
may increase wettability and, consequently, reduce foamability. SiC-reinforced
wrought alloys cannot be remelted without severe aluminum carbide formation
at the interfaces. The formation of Al4C3 would be detrimental as it significantly
reduces the corrosion resistance of the component. Furthermore, it affects the
foamability by increasing the viscosity of the melt and above all by providing wettability. Nonwettability of the additives is necessary for the trapping of gas bubbles.
In the case of Al casting alloys containing 7 12 wt.-% Si reinforced by SiC, there is
the chance to conserve the integrity of SiC by keeping the melt temperature below
750 hC as recommended for primary foundry technology [13].
A detailed study of the recycling of SiC-particle reinforced Al Si casting alloys is
given elsewhere [14], where the quality criteria of the melt are given and remelting,
recycling, and holding practices are described.
x
x
x
x
x
x
x
Dry, pre-heated scrap can be added to the melt between 700 and 750 hC.
During a rest period, usually more dross is formed than on primary material. It
contains mainly oxides and SiC-particles without reducing significantly their content in the melt.
The dross has to be skimmed.
An impeller is introduced to produce strong movement under the surface skin.
Fluxing and degassing with argon (SF6 may also be used) to remove oxides and
reduce the hydrogen content.
The melt is allowed to sit and then skimmed.
The melt is mechanically agitated, without forming a vortex, to distribute the SiC
particles homogeneously.
No loss in particle content was reported and the removal of porosity, oxide films,
and hydrogen were efficient. The amount of dross generated is relatively high
and may amount to more than 10 % of the total weight.
If the ceramic particles should be removed from the aluminum melt, conventional salt addition, or fluxing techniques (as are executed to remove oxide films
[14]) can be applied. Gravity settling allows the fluxed ceramic ingredients to
float to the dross at the top of the aluminum melt. Rotary salt furnace technology
is an established reclamation process to recover aluminum from various mixtures,
including particle reinforced metals; however, this requires 20 50 wt.-% salt [13].
Both wrought and foundry alloys and even machining chips can be recovered
using this technique. Recovery of about 80 % of the available aluminum can be expected [15].
The efficiency of particle removal by fluxing is related to the probability of contact between the ceramic constituent and the flux. Duralcan, a supplier of particulate-reinforced aluminum, proposes to incorporate the salt into a melt agitated by
gas injection [16]. Thus dewetting is achieved with much smaller salt additions
(I1 wt.-% for alumina and about 1.5 wt.-% for SiC) and in combination with
31
32
Conclusions
There are the following possibilities for recycling, or reuse of cellular metals (especially cellular aluminum).
x
Recycling of unreinforced cellular metals can be carried out as for packaging material (foils), or shredder scrap, by remelting, with an efficiency probably reduced
to 80 % to produce bulk cast products from the secondary metal. In the case of
recycling ALPORAS, the impurity level of Ca has to be controlled.
Recycling of MMC foams can be done by extracting the matrix metal for conventional secondary products. The reinforcement may be segregated to the dross and
deposited.
Recycling of MMC foams can alternatively be achieved by remelting according to
the precautions for MMC recycling, but at a reduced efficiency due to the increased oxide content to produce MMC foams again.
The foam production itself may be based on an up-grading of secondary material
with higher contents of impurities that increase foamability. In particular, the
cycles for MMC and MMC foams can be closely related.
References
2.5
A foam is a dispersion of gas bubbles in a liquid in which the bubbles are deformed due to their mutual interaction. Cellular metals produced via gas bubbles
in liquid metal are commonly named metal foams although they do not strictly
meet the definition above after solidification. Owing to the surface energy necessary to form the metalgas interface a foam is never in equilibrium and hence permanently trying to lower the internal energy by reducing the internal surface. That
is, the cellular structural state of a foam evolves with time and its actual structure is
a function of its history including all thermal and mechanical influences. The foam
structure is normally strongly disordered and evolves by some combination of
three basic mechanisms: bubble coalescence via film rupture; bubble coarsening
via diffusion of gas from smaller to larger bubbles; and drainage downwards
and out from the foam in response to gravity [1]. It is only in recent years that
some progress has been made towards an understanding of the basic mechanisms
governing the temporal evolution of foams [15].
Owing to their opacity and the high temperatures, in-situ observation of the
structure evolution of metal foams is difficult [6]. Fortunately, the solidification process is in general much faster than the evolution processes at least as far as the
evolution is not caused by expansion. That is, nearly full information of the
foam formation process can be extracted from ex-situ investigation of foamed samples in different stages of expansion.
33
34
2.5.1
R_
3 _2
1
2s
(Pbubble s
s PT )
R S 4v
R
2
r
R
(1)
where Pbubble bubble pressure, PT equilibrium pressure in the liquid, s surface energy, n kinematic viscosity, r density.
The left hand side of Eq. (1) describes the inertia and viscous forces that both
delay bubble growth. Neglecting viscous effects the time for the formation of a bubble in aluminum with radius R 1 mm at a bubble overpressure of DP 10 4 bar
is given by [7]
r
r
(2)
0:015s
t 0:915 R
DP
That is, inertia effects delaying bubble expansion can be neglected if foam formation is on a time scale of seconds. The contribution of viscous forces for liquid metals is normally very small, for example about 4nr(R/R) z 10 6 bar with R 1 mm/
s; R 100 mm; n 1 mm2/s. On the other hand, if foaming takes place in the
semi-liquid state, where the viscosity of the metal is several orders of magnitude
higher than in the liquid state, viscous forces might delay bubble expansion.
Bubble expansion takes generally more than one second for commercially known
foaming methods for metals [9]. In this case, the viscous and inertia forces can be
neglected and the Rayleigh equation reduces to pressure equilibrium at the gasliquid interface
Pbubble PT S 2
s
s
P0 S rgh S 2
R
R
(3)
For a pure aluminum melt and bubble radii R of 100 mm and 10 mm the rising
velocity is about 1 cm/s and 100 mm/s, respectively. Additives in the fluid like SiC
or Al2O3 particles influence the movement of the bubbles and are able to stabilize
them [1012]. They have an effect on both the viscosity of the melt and the surface
tension. How these particles actually operate and how their action can be optimized is not yet understood and still a matter of research.
A gas bubble grows or shrinks due to gas exchange with the surrounding melt
[13]. There is a gas flow from the liquid into the bubble if the concentration of dissolved gas in the liquid, as a result of the decomposition of a foaming agent for
example, is higher than the equilibrium concentration in the liquid given by
Henry's law for a given gas pressure in the bubble. Since hydrogen, which is preferentially used as foaming gas, dissociates when dissolved in aluminum the equilibrium hydrogen concentration, ceq, at the gasliquid interface is governed by
Sievert's law, a special form of Henry's law [14]
ceq 1:4 q 10s3 q 10s
2760
T
p
Pbubble
mol
p
cm3 bar
(5)
35
36
Figure 2.5-3.
DGhom
kT
with DGhom
16ps 3
3DP2
(6)
DGhet
kT
(7)
37
38
cell walls are curved due to the pressure difference of neighboring cells. One effective way to characterize the cell structure is by the shape factor, F, calculated from a
2D cut through the foam [23] (see also Section 4.1)
F
n
4p X
ai
n i li2
(8)
Figure 2.5-5.
last expression is deduced from a simple cubic plate model for the cells with the
assumption that the volume of the cell-forming material per cell is constant. Actually, the experimental data (see Fig. 2.5-7) is much better fitted by
D
1s
d
3d
p
for rrel II 1
3
1 s rrel
rrel
(9)
which follows from a cubic plate model assuming a constant mean cell-wall thickness, d. That is, the mean cell-wall thickness is constant during expansion. It depends on the alloy and is found between 50 mm and 300 mm for aluminum
foams. Consequently, given a fixed alloy cell size and relative density are intimately
related and can hardly be influenced by processing parameters [23]. In order to
generate aluminum foams with a lower density and smaller cells the mean cellwall thickness has to be reduced. How the cell walls of metal foams are stabilized
by the particles and which properties of them (quantity, size, form) determine the
mean cell-wall thickness is not yet understood and a matter of flow research [18].
Coalescence, rupture of a cell wall, is a mechanical process that depends on the
mechanical stability of the membrane. It has statistical character and occurs if a
local fluctuation of the film thickness leads to faster local thinning because it is
not compensated by restoring forces. The different physical mechanisms that are
responsible for the mechanical stability of a membrane have been extensively investigated for aqueous films [5,15]. Experimental studies of the stability of metal
membranes are not known in literature. Physical forces between the two surfaces
of the film like the van der Waals or electromagnetic interaction are only relevant
39
40
for very thin films (I1 mm). For thicker films, as it is the case for metal foams,
there are two effects that might lead to a kind of elasticity of the membrane.
The Gibbs' effect [15] takes into account that a local thinning of the membrane
leads to a local thinning of the surfactants. As a result, the surface energy increases
at this point leading to a restoring force. The Marangoni effect [15] describes that a
gradient of the surface energy leads to a flow of surfactants taking with it a thin
film of fluid in the direction of lower concentration. In order to maximize both effects the surfactant should minimize the surface energy. At the moment, it is not
clear whether the stability of metal membranes is based on one of these two mechanisms. In metals, melt surfaces are covered with oxides and other particles that
might have the required effects on the surface energy.
The origin for cell-wall instability is cell-wall thinning, which can be traced back
to different mechanisms. The first one is simply related to foam expansion itself. If
a cell expands the cell-wall thickness decreases due to mass conservation. During
foam expansion the principal mechanism for cell-wall thinning is by growth. The
second mechanism of cell-wall thinning is drainage due to gravity and capillary
forces. Drainage determines the foam stability in the absence of further expansion
due to gas release, that is, the time that a given foam structure is maintained without destruction. Gravity induces a melt flow from the top to the bottom of the foam
and generates a density gradient [4,5]. Neglecting capillary forces, the draining velocity, vD, can be estimated in a first approximation by the balance of gravity and
viscous stress from shear flow within the Plateau borders [1]
vD Z
grd2edge
h
(10)
That is, the draining velocity increases with the cell-edge thickness, dedge.
Capillary forces arise due to the different curvature of the interfaces. They force
the melt from the cell membranes to the cell edges (Fig. 2.5-8). The rate of cell-wall
thinning due to this pressure gradient can be modeled using a Reynolds-type
equation for the flow between two circular parallel discs with immobile wall surfaces [5]
x_ s
2sx 3
3hR2 RPB
(11)
Figure 2.5-8.
41
42
meter. As a result, pores are deformed and also destroyed during mold filling by
mechanical forces or a non-uniform foaming velocity (Fig. 2.5-10). Hence, the
internal foam structure is strongly influenced by the particular production procedure.
References
2.6
The process of replicating a structure has been known to the metallurgist for centuries. For example, casting can be thought of as a replication procedure in which a
metal is used to reproduce the negative form of the mold. In the study of porous
materials, however, replication refers to a process used to replicate the open-pore
architecture of a porous material.
43
44
2.6 Infiltration and the Replication Process for Producing Metal Sponges
2.6.1
Replication
More specifically, this process can be defined as a general three-step procedure for
the production of highly porous materials (Fig. 2.6-1). The three steps are:
1. preparation of a removable pattern;
2. infiltration of the pattern followed by solidification (or an alternative process,
such as curing or cross-linking, which rigidifies the infiltrate); and
3. removal of the pattern.
In some cases, a fourth step, such as pyrolysis, may be required to transform the
porous network into the desired phase. In this text, we focus on the use of the replication process for the production of cellular metals with open porosity. Several
Figure 2.6-1.
authors have noted that metal (or metallic) sponge is perhaps the most appropriate
terminology for these materials as both the term foam and cellular imply a
closed-cell architecture [1,2], while porous metals produced by replication must
have an open-pore network. We will thus refer to metals with open porosity of
any density I0.8, produced by replication, as metal sponge.
This process is distinct from deposition techniques, such as those used by Inco
Limited (Swansea SA6 5QR, Wales, UK) to produce porous nickel substrates, called
Incofoam [3]. In that process, a temporary pattern is used, but the pore space is not
replicated: instead, the pattern is coated and the pore space only partially filled.
Sintering methods that utilize a space holder are also not considered replication
for two reasons: the micro-architectural features of the pores are determined by
the metal powder size in addition to the space-holding material and infiltration
is not one of the processing steps. The so-called lattice block materials (LBM)
are prepared using polymer patterns of truss structures and investment techniques; thus the lattice block is rather an engineered structure and cannot really
be considered a metal sponge, which can be shaped and formed while maintaining the statistics of its properties, as long as the cell sizes are of millimeter
or more dimensions. For more details on lattice block materials contact JAMCORP
(17 Jonspin Road, Wilmington, MA 01887-1020 USA). Although we limit this discussion to metals, we note in passing that the polymeric sponge technique used to
produce porous ceramics also does not replicate the pattern and should not be considered a replication process.
In a metallurgical context, the replication technique can be used to produce
metal sponge that is difficult or impossible to produce by other processing methodologies. Replicated metal sponge, for example, by necessity has an open-pore
structure, which is a fundamental difference compared to most commercially available foamed metals. The structure or architecture of metal sponge produced by replication is flexible and determined by the pattern: porosity as high as 98 % [4] and
as low as 55 % [2] for different pattern materials have been reported and pore sizes
as small as 10 mm have been achieved [5]. In addition, metal sponge can be
produced from virtually any alloy that can be cast (provided a suitable pattern material exists). The replication process has been used extensively for the production
of various porous materials including carbon [6], silicon carbide [7], aluminum alloys [2,4,5,813], magnesium alloys [14,15] and alloys based on iron and nickel
[4,16].
In the following sections we provide details about the production of metal
sponges by replication techniques. We first describe the important physical phenomena that govern each of the three basic steps of the replication process (with
some emphasis on sponge produced from salt patterns for which the most information exists in the literature). We then provide some physical and mechanical
properties of metal sponge produced in this manner.
45
46
2.6 Infiltration and the Replication Process for Producing Metal Sponges
2.6.2
is determined by the size of the initial spheres (or particles) and the ratio of polystyrene to resin. This type of pattern has the advantage that it is not necessary
to physically remove solid grains from the structure after infiltration, as the pattern
is essentially burned away at the end of the process. On the other hand, this pattern must be infiltrated relatively cold to avoid burning of the pattern during infiltration (i. e., while the metal is still liquid) and is thus limited to low-melting alloys
such as aluminum, magnesium, and zinc.
The fourth class of pattern material is leachable: these patterns are removed by
dissolution in an appropriate solvent. The process of dissolution (or leaching) is
less restrictive on size and density than insoluble refractory materials. In practice
however, patterns of leachable material are created from granules or powder, thus
the density is related to the packing efficiency of the granules, which itself depends
on the size (and size distribution) of the granules.
Sodium chloride (NaCl) has been employed most extensively as a leachable pattern, because it is inexpensive and easy to handle. Sodium chloride patterns used
for the production of porous aluminum were first reported [8,9] and patented [10]
in the 1960s, and have advanced more recently [5] (there are also reports of their
use for carbon sponge [6] and for SiC sponge [7], both with a pyrolysis step).
Salt patterns are limited by their melting point; NaCl, for example, is limited to
aluminum and lower-melting alloys, while NaF could potentially be used at temperatures greater than 900 hC. Additionally, highly concentrated saline solutions
are generated during dissolution and can cause significant corrosion in some alloys.
Salt grains also have the advantage that they can be sintered to enhance the connectivity of the salt, change the structure of the pattern, and create a free standing
preform that can be handled and subsequently infiltrated [57]. The sintering step
is not necessary, however, for two reasons: transient bonding of the grains (perhaps
due to humidity in the salt [8]) may occur during preheating the salt prior to and
during infiltration, and incomplete penetration of the melt locally where individual
grains meet, provides the necessary open-network for salt removal.
Sintering processes depend on a number of parameters, including particle size,
temperature, atmosphere, residual stresses, and time. Such numerous process
parameters might explain some of the varying observations in the literature with
regard to the sintering of sodium chloride. Incidentally, these varying observations
point to the fact that sodium chloride may be a good choice for a pattern material
because of flexibility in controlling the sintering mechanisms and thus the microarchitectural features of the pattern.
There is a consensus in the literature that large sodium chloride particles sinter
predominantly by an evaporationcondensation mechanism, in which the necks
between the particles grow without densification of the salt compact. Small particle
compacts, on the other hand, can densify significantly during sintering, due to the
predominance of other mechanisms where the center-to-center distances between
particles can decrease, such as bulk diffusion [1821]. The minimum particle size
that can sinter without densification has been estimated to be about 150 mm
[19,20]. Sintering in vacuum may promote the evaporationcondensation mechan-
47
48
2.6 Infiltration and the Replication Process for Producing Metal Sponges
ism for smaller particles [22] as supported by the pressure dependence that was
noted for the rate of neck growth when non-densifying sintering mechanisms
are dominant [23].
There are advantages to both classes of sintering mechanisms with respect to
producing metal sponge. Non-densifying mechanisms, such as evaporationcondensation should develop a structure dictated by capillary equilibrium: these are
well known [24]. The pore architecture in such a case is expected to be similar
to the regular structures of polymeric sponge. Densifying mechanisms, on the
other hand, provide a means to achieve relative densities significantly greater
(and thus significantly lower for the sponge) than those attainable by packing
and pressing of particles. Sintered salt densities as high as 90 % have been reported
with nearly all the pores as open porosity, although for a particle size of 1 mm,
which would pose challenges in the subsequent processing steps [25].
Infiltration
Ideally, infiltration under the influence of gravity is preferred as it does not require
special equipment or procedures. In most cases, however, the metal does not wet
the pattern and some additional force must be applied to the melt to promote (or
assist) infiltration and effect a uniform distribution of metal within the pattern.
Vacuum and low-pressure assisted casting procedures are standard foundry technologies and, in many cases, these are sufficient for infiltration. Indeed, it has
been reported that spontaneous infiltration is possible when infiltrating salt grains
greater than 4 mm in size. Vibration slightly reduces the size of salt grains that can
be infiltrated under gravity, while vacuum applied to the pattern with a slight overpressure on the melt has also been used to infiltrate grains as small as a few hundred micrometers in diameter [8].
The specific pressure required to initiate infiltration (the threshold pressure, P0)
depends to a large extent on the size of the pores to be infiltrated and the volume
fraction of these pores. A detailed physical description of infiltration is beyond the
scope of this work; more detailed information is available elsewhere [2628]. To the
authors' knowledge, the pressure required to initiate infiltration has not been explored with specific emphasis on patterns used for replication processes; however,
expressions derived for the threshold pressure in the infiltration of packed ceramic
beds with pure aluminum should be directly relevant. Garcia-Cordovilla et al. offer
a simple semi-empirical relationship [27]
2.6.2.2
P0 MPa z 16
f
s 0:09
(1 s f )D
where f is the volume fraction solid (or in this context the fractional density of the
pattern) and D is a characteristic diameter of the solid in micrometers. This relationship is based on data for particles in the range 10100 mm packed to a density
of 5060 % and infiltrated with pure aluminum, but it should provide reasonable
magnitude estimates of the pressures required for infiltration over a broader range
Pattern Removal
The pattern removal step depends on the characteristics of the pattern. Patterns
from fine-grained ceramic powders such as investment casting compound and agglomerated silica pellets are typically removed by spraying water on the pattern
metal composite [1216], such that the water physically breaks down the pattern
into the small grains from which it is made. This requires penetration of the porous network by the spray, or at least that the fluid can wick into the grain structure of the pattern so that it breaks down. For relatively fine (I1 mm) and dense
sponges i10 %), however, it is unclear how efficient this process would be for sections many times the pore size. Additional methods such as (ultrasonic) vibration
may also prove useful for removal of solid grains.
2.6.2.3
49
50
2.6 Infiltration and the Replication Process for Producing Metal Sponges
Polymer pattern materials can be removed by burning in air as is the case for
polystyrene-resin patterns [11]. The cleanliness of this burning is an important
issue in an industrial setting. These patterns, however, have the advantage that
complex parts can be machined before removal of the pattern and without significant damage to the structure of the sponge.
In some cases, leachable patterns can also be machined to complex geometries
prior to leaching and without damaging the architecture of the sponge. In the case
of salt, removal of the pattern is by immersion in water, causing dissolution of the
salt. Dissolution is primarily a diffusive process, as the dissolved salt ions diffuse
from their place-holding position into the water bath through the narrow channels
in the sponge where the water is essentially stagnant. Thus, the rate and total time
of dissolution depend strongly on the size of the specimen. Salt is more difficult to
remove from small pores, particularly if gas is entrapped in the pores; this is the
so-called gas lock phenomenon, which can block the penetration of fluid into the
porous network [8].
Corrosion of metal sponge during leaching of the pattern can be a problem due
to the large surface area of the porous network, thus the salt concentration and the
immersion time must be minimized. This is accomplished by simply changing the
water periodically, however, it is important that the sponge is not drained at this
point, as this will cause gas to be entrapped in pores and promote gas lock. Additionally, impurities in the water should be controlled to avoid precipitation of
other salts in these concentrated solutions. Aluminum alloys (the metals for
which salt patterns are most applicable) tend to be relatively corrosion-resistant
in salt solutions. The aluminumcopper alloys are among the least corrosion-resistant in salt solutions, but sponge can still be prepared from these alloys without
degradation by corrosion. Electrochemical protection could also offer additional
protection from corrosion.
A large volume of water is not necessary as salt-saturated water is denser than
distilled water, and hence falls to the bottom of the container where it remains
in the absence of convection. As a result, during the dissolution process a large
concentration gradient is maintained between the outer surface of the saltmetal
composite suspended near the surface of the dissolution bath (but completely immersed) and the bottom of the container. For large pores and low-density sponge,
mixing may modestly enhance dissolution if the velocity of the fluid in the porous
network is affected. In most cases, however, this condition is not met and mixing
the water primarily distributes the salt concentration and reduces the diffusion gradient in the water bath, in turn slowing the leaching process.
The final stage of salt removal from the sponge is rinsing. This can be accomplished by flowing water through the structure, or simply by changing the water
several times to incrementally remove the saturated fluid; this process is greatly
facilitated by the higher density of salt solutions. Pressurized gas is also effective
for removing the residual fluid in the sponge such that dissolved salts are not
left in the structure when this water evaporates.
Other methods may contribute to more efficient (faster) removal of salt, but
these must enhance the rate of diffusion of the salt in the fluid. For example, ul-
trasonic vibration has been proposed as a useful method as it should increase the
rate of diffusion with the added benefit of freeing gas bubbles entrapped in the porous network [8].
2.6.3
Published reports of metal sponge produced by replication techniques are summarized in Table 2.6-1. Relevant details of the processes and properties of the resulting
metal sponge are briefly described below in the context of the basic pattern technologies: continuous refractory, discontinuous refractory, burnable, and leachable.
51
Al
(SG91A)
pig Fe
Al
Al
NaCl
Al alloys
NaCl (presintered)
Mg
(AZ91)
Al
Metal
gas-pressure infiltration
vacuum assisted
vacuum or pressure
assisted
vacuum assisted
vacuum assisted
Infiltration technique
leaching
leaching
burning
pressurized water
(not reported)
water spray
water spray
Pattern
removal
Pattern material
Table 2.6-1.
0.013
0.451.45
0.155
1.23.7
15
0.852.5
4.5
4.5
6581
5567
7386
6063
9298
9396
97
Porosity (%)
[5]
[2]
[8]
[11]
[15]
[4]
[12, 13]
[14, 15]
Reference
52
2.6 Infiltration and the Replication Process for Producing Metal Sponges
Burnable Patterns
Aluminum sponge has been prepared from resin-bonded polystyrene spheres
(Table 2.6-1) with porosity in the range 7386 % and pore sizes in the range
14 mm [11]. Infiltration was conducted with a modified die-casting technique
and the pattern was removed by burning in air. Properties of these metal sponges
have not been reported.
2.6.3.3
Leachable Patterns
Aluminum sponge was produced from NaCl as early as 1961 [8]. The patterns were
prepared by packing salt grains in a mold and infiltrating by one of several techniques depending on the size of the salt grains: by gravity (i 3.4 mm), on a vibrating table (i 1.7 mm), or with an overpressure and simultaneous application of
pressure (j 0.15 mm). The compressive strength of an A356 aluminum alloy
sponge was observed to depend on the heat treatment and at 10 % strain varied
in the range 2030 MPa for porosity in the range 6063 %. The porosity undoubtedly varied depending on the size (and distribution of sizes) of the salt grains. In
an independent study, Banhart reports a range in porosity between 55 and 67 %;
however, he does not provide any details about the processing other than that
the pattern was leachable [2]. Some basic property indicators (mechanical response
to compression, pressure drop and sound adsorption) are given in this latter report.
Detailed analysis of properties is not given in either reference.
Recently, gas-pressure infiltration techniques have been used to infiltrate
sintered NaCl with grain sizes as small as 10 mm [5]. Porosity was in the range
6881 % depending, in part, on the size of the salt grains. A SEM micrograph
of an aluminum sponge prepared by replication of sintered NaCl is shown in
Fig. 2.6-2. The mechanical response of a metal sponge depends on the microstructure of the metal (Fig. 2.6-3), as with any fully dense alloy; however, the properties
of the metal sponge are further influenced by the level of porosity. The shape
of the flow curves in Fig. 2.6-3 are somewhat different from what is generally
observed with commercially available closed-cell metal foams. This is due to
the inherently higher ductility of most of the base metals used to produce the
sponges represented in Fig. 2.6-3 in comparison to commercial cellular metals,
as well as the higher regularity of the architecture of these metal sponges, and
the comparatively low porosity. The stiffness was found to be about half that
expected from simple phenomenological relationships [5].
The processing limits for replication from leachable patterns, like the other pattern technologies, have not been clearly evaluated and there is some divergence in
the literature. The range of available pore sizes and porosity, however, is clearly
considerable.
2.6.3.4
53
54
2.6 Infiltration and the Replication Process for Producing Metal Sponges
Figure 2.6-2. Scanning electron micrograph of aluminum sponge produced by replication from
sintered NaCl patterns.
2.6.4
Conclusions
Replication processing provides a versatile and attractive method for the production of metal sponge (often called open-celled metal foam). Advantages of this process comprise considerable control of the pore size, density, and internal architecture of the sponge, the capacity for net or near-net shape component production,
and the potential for production of components that are part dense, part porous.
Also, the open-porosity that is intrinsic to metal sponge can be put to advantage
in several applications that require fluid flow through the porous metal, such as
filters and heat-exchangers.
The replication process exists in a few variants, which differ mostly according to
the nature of the pattern: continuous refractory, discontinuous refractory, burnable,
and leachable patterns are four classes of patterns that have been employed to date.
The range of density and pore size that can be achieved in replicated metal sponges
depends largely on the pattern material and also on the infiltration process. With
leachable patterns, in particular, these ranges have been shown to be comparatively
wide; at present these also seem to be complementary to what can be achieved with
the methods utilizing insoluble patterns. There is, though, significant room for innovation in replication technology as related to the production of metal sponge and
there exists perhaps even opportunities for cross-pollination between the methods
described here.
References
55
56
2.7
In this chapter, solid-state and deposition methods for the manufacturing of cellular metals are reviewed with regard to their most important processing features.
Discussion will be limited to methods producing predominantly 3D structures.
Typical values of the cell size, cell wall thickness, and obtainable relative densities
are cited wherever available. Earlier overviews including both methods of manufacturing are available [1,2].
The term solid-state method means that directly before the formation of the
cellular structure, the metal is in the solid state. However, in some cases limited
amounts of liquid phase appear in the process: during liquid-phase sintering or
brazing of single cells. Except for foaming of solids, all solid-state methods require
a sintering step before the final cellular metallic structure is obtained. Naturally,
powder metallurgical methods play a dominant role in this area.
Porosity in cellular structures made from powders often exhibits two different
size scales: a macroporosity of the order of the targeted cell size, and a microporosity in the order of the primary particle or, in case of reactive sintering, the
grain size.
For all powder metallurgical processes, the cost of the powders is a major concern. Some of the methods, especially those using slurries of metal powders dispersed in a dissolved binder, are capable of using cheap oxides and thus have
the potential for a significant cost reduction.
Deposition methods need to mobilize the metal in order to transport it onto a
porous substrate. This is normally done by transforming the metal into the ionic
or vapor state. From the method of manufacture, it can be deduced that the deposition methods are only capable of generating an open-cell morphology. Usually, they
also imply a heat treatment for the removal of the precursor on which the metal is
deposited.
A practical differentiation between the methods in question can be made with the
help of two manufacturing parameters, as is shown in Fig. 2.7-1. The chosen criteria,
formation of cellular structure and formation of cells have direct implications for
the processing itself, as well as for the resulting properties of the structure.
According to this classification, the formation of the cellular structure can be accomplished by building it up from single cells, which do not necessarily have to be
in the final metallic state. Alternatively, the structure can be built up from bulk.
The formation of the cells themselves can either be carried out with the help of
a lost core, or in a coreless manner. Within this framework the deposition methods, for example, constitute a small subset of the lost core bulk processes.
Solid state
and deposition methods
from single cell
from bulk
coreless methods
structures from hollow
powders
foaming of solids
spraying of slurries
foaming of slurries
Classification of solid-state
and deposition methods from a manufacturing point of view.
Figure 2.7-1.
cementation process
deposition methods
57
58
Generally, formation from single cells has the potential of a very close control of
the cellular properties. However, in most cases the single cell processes require
more processing steps than bulk formation processes.
Processes using lost cores usually are prone to introducing unwanted interstitials
like carbon and oxygen into the base material, since the cores are predominantly hydrocarbons, which have to be removed thermally. Another disadvantage compared to
coreless processes is the additional time required for the removal of the core.
Generally speaking, there is a remarkable variety of solid-state and deposition
methods for making cellular metals, which all have their specific strengths and
weaknesses. It therefore depends very much on the final application, which
method may be best suited in a given situation.
2.7.1
a) Surface
b) Cross section
1.0 mm
1.0 mm
59
60
achieved with stainless steel structures made from reduced iron and chromium
oxide powders [9], Fig. 2.7-3.
2.7.2
Cu2 Fe () Cu Fe2
leads to deposition of pure copper on the particles while simultaneously the iron
core is being dissolved. The process can be carried out until the iron is completely
dissolved in the electrolyte and only hollow copper shells remain [10].
Hollow copper powders were prepared using reduced iron oxide powders as the
lost core. The resulting copper particles had diameters of about 500750 mm. The
walls are very rough and porous, having a thickness of a few hundred microns. The
hollow copper particles need further heat treatment to give them sufficient
strength for handling. Cellular metallic structures are achieved by packing the hollow particles in ceramic molds and bonding them by sintering.
Good results were achieved by coating the copper particles with tin, thus enabling liquid-phase sintering. The resulting cellular bronze structures exhibit
good interparticle bonding, dense cell walls, and relative densities of about 0.2
[11], Fig. 2.7-4.
Cross section of a bronze hollow-sphere structure made by cementation and sintering (courtesy Institute for Chemical Technology of Inorganic Materials, Technical University of
Vienna).
Figure 2.7-4.
Figure 2.7-5. Brazed nickel hollow-sphere structures with sphere diameter of 4.5 mm made from
galvanically manufactured hollow spheres (courtesy ATECA, Montauban, France).
61
62
are heat treated in the same way as the single spheres. In principle, this allows for
the production of hollow-sphere structures from arbitrary metals and alloys.
In order to enhance the mechanical properties of the resulting structures, large
contact areas between the spheres are achieved by applying a slight pressure to the
green spheres during the forming step. In this stage, the elastic Styrofoam core
helps to prevent buckling and leads to a more uniform distribution of the cell
pre-deformation. A typical example of the resulting structure is shown in
Fig. 2.7-7.
The attainable range of the sphere diameters is 0.510 mm. Wall thicknesses
have been made from 10 to 1000 mm. Relative densities can be as low as 0.03. Recent overviews of the processing and properties of such structures are given elsewhere [17,18].
Materials currently under investigation are numerous, with a focus on stainless
steel 316L [19] and low-cost hollow-sphere structures made from iron oxides [20].
2.7.3
Figure 2.7-8. Cross-sectional cut of sintered fiber structure made from melt-extracted short
fibers from FeCrAl 23.15. From left to right: relative density 0.3, 0.2, 0.1 (courtesy IFAM,
Department of Powder Metallurgy and Composite Materials, Dresden).
63
64
Using the melt extraction process for making short fibers directly from the melt,
the materials limitations of the mechanical fiber manufacturing routes can be overcome [24,25]. Highly porous sintered structures from intermetallics such as Ni3Al
[26] and FeAl, as well as from many other ferrous and nonferrous alloys have been
made. Equivalent fiber diameters are 50150 mm, fiber length may vary between
3 and 25 mm. This allows for relative densities between 0.5 and 0.03 in the final
structure, Fig. 2.7-8.
Porous TiAl3 structure prepared from elemental powder mixture. Large pores
correspond to primary Al particles (courtesy IFAM, Department of Powder Metallurgy and
Composites).
Figure 2.7-9.
Foaming of Solids
Kearns et al. first described a solid-state foaming process where porosity in metals
is generated by expanding a hot-isostatically pressed mixture of argon gas and
metal powder at elevated temperatures and ambient pressure [31].
More recent work utilizes this principle for making TiAlV 64 sandwiches with
a porous core [32]. This is achieved by canning TiAlV 64 powder in canisters of
the same material. After HIP, the resulting billet can be isothermally forged or
hot rolled to achieve sheets. Subsequently, an expansion annealing treatment
(typically at temperatures i0.6Tm for 448 h) generates the porous core,
Fig. 2.7-10.
This method, which is known as the low-density core (LDC) process, uses creep
expansion due to the internal gas pore pressure to generate the desired porosity.
Queheillalt et al. have carried out a thorough analysis of the process kinetics
and morphological evolution of the pores [33]. The resulting porosity in the core
region is largely unconnected and shows pore diameters of 10100 mm. The relative density lies in the range 0.650.5. One advantage is the capability of producing
very large sheets (up to 2000 mm q 1200 mm q 4 mm).
A slightly different approach described previously utilizes the transformation
superplasticity found in allotropic materials [34]. Transformation superplasticity occurs at all grain sizes by biasing with a deviatoric stress (from the entrapped gas) of
internal stresses (from the allotropic mismatch during thermal cycling about the
allotropic temperature range). Hence, the rather lengthy expansion annealing
can be replaced by a significantly shorter thermal cycling treatment.
2.7.3.3
65
66
Ti-6Al-4V can
Powder
packing
density D0
Ti-6Al-4V
powder
b) HIP Consolidation
(1040oC, 100 - 200MPa, 2h)
Isolated
pressurized
voids
p(t), T(t)
c) Hot Rolling
(approx. 850oC, 6-20 passes)
Thinning of
facesheet
Internal gas
pressure
Changes in
pore shape,
matrix
microstructure
Isolated
porosity
(<40%)
Facesheet
Figure 2.7-10.
Foaming of Slurries
There are only few reports on the foaming of slurries, and work has not been very
successful up to date owing to low strengths of the resulting structures. According
to Banhart et al., metal powder slurries containing a foaming agent can be poured
into molds and foaming occurs due to decomposition of the foaming agent [2]. Stabilizing the resulting liquid foam is essential and in some cases is accomplished by
adding a stabilizer capable of polymerization on contact [1]. After foaming, the organic contents have to be removed thermally, followed by a sintering step. Relative
densities as low as 0.05 have been reported [35].
2.7.3.4
2.7.4
Figure 2.7-11. Titanium sample with relative density of 0.3 made by space holder method
(courtesy Research Centre Ju
lich, Institute for Materials and Processes in Energy Systems).
67
68
Deposition Methods
Deposition techniques always need a parent structure on which to deposit the
metal. Most common is the use of polymeric (polyurethane) foams. In order to ensure the penetration of the structure by the mobilized metal, a thermophysical
shock treatment (reticulation) for the removal of residual membranes is needed.
After deposition, the parent structure is removed thermally or chemically.
If the metal is deposited from the ionic state (either galvanically or electroless), a
pretreatment is necessary that assures either electrical conductivity or, in the case
of electroless plating, a sufficiently rough surface of the parent structure.
Materials choice is limited to nickel, nickelchromium, silver, copper, and tin.
Relative densities are typically less than 0.1 with average cell sizes from 400 mm
up to a few millimeters.
In order to overcome the limitations in materials choice, it is also possible to
deposit the metal from the vapor state. Albeit being a nickel foam, INCOFOAM
made by Inco Ltd., Canada, is an example of a commercially available product
made by chemical vapor deposition. A recent development in this area is the electron beam directed vapor deposition process (EB-DVD), which was used to evaporate and deposit an Inconel alloy 625 onto an open celled, reticulated structure
[38], Fig. 2.7-12. The EB-DVD technology uses a combination of an electron
beam gun and an inert gas carrier jet of controlled composition to create thick
coatings with high deposition rates (i10 mm/min) and could thus overcome
the economical limitations encountered to date by vapor deposition methods.
2.7.4.2
References
69
70
3
Secondary Treatment of Cellular Metals
F. Simancik
The introduction of new materials into practice requires suitable technologies for
their machining, forming, surface coating, or joining. Many papers in the literature
deal with the preparation of cellular metals and with the evaluation of their mechanical and physical properties, but only limited attention is paid to their secondary treatment. There is no doubt that the following unusual features of cellular metals require specific approaches for shaping and fabrication.
x
x
x
x
71
72
has to be taken into account, especially if SiC particles are present. Conventional
machining causes bending of thin pore walls in ductile alloys and fracture in brittle
alloys. Pore walls yield or fracture at their thinnest point, which is often out of the
cutting plane. This leads to poor quality of the cutting surface and low precision.
The accuracy and quality of the cutting plane are significantly affected by partial
melting of the thin pore walls and their subsequent diffusion bonding to the cutting tool. High porosity and the resultant low thermal conductivity do not allow for
efficient cooling by transport of heat into the surroundings. On the other hand,
melting and bending of pore walls can be utilized for the strengthening of the cutting surface (bent pore walls are pressed into the surface pores, which improves
their apparent density).
If high precision and a good quality machined surface is required, traditional
methods of machining cannot be used. Electric discharge machining was demonstrated to be most suitable, as it avoids the mechanical effect on the pore walls and
thus their bending and failure. Material in the cutting plane simply burns out. The
result is a clean cutting plane (without material pressed into the surface pores) and
good dimensional accuracy. The speed of electric discharge machining is substantially greater in porous materials than with dense ones. The main disadvantage of
the process is the need to use dielectric fluid, which penetrates into the porous
structure and needs to be removed after machining by burning: this is environmentally harmful. If a simple cut is required, wire electric discharge machining
in water can be used to overcome this disadvantage. A water beam with abrasive
can also be used for the precise cutting of cellular metals. The quality of the cutting
plane is substantially better than in the case of classical band saws (the bending of
pore walls does not take place). As the cutting speed and the thickness of the plate
increases, the quality of the bottom edge of the cutting plane deteriorates owing to
scatter of the beam. The same effect takes place in plasma or laser cutting.
If the component is foamed, its final shape derives from the mold and forming
is usually not necessary. From the practical point of view, forming can be considered with respect to flat plates. Sometimes bending or roll-bending of these panels
is required. During bending, cracks initiate in the tensile surface and then they rapidly propagate across the whole section. The bending that can be achieved without
damage is limited by the plasticity of the thin surface skin and the thickness of the
plate (as the thickness of the plate grows, the deformation of the tensile loaded part
of the sample for a given bending radius increases). Panel bending is therefore possible only for foams made of ductile alloys and/or at higher temperatures (see
Section 3.1).
Surface coating
For surface coating the same basic principles can be applied as for the bulk cellwall material, but soaking of the porous structure should be taken into account
as it significantly restricts those surface treatment procedures in which components are immersed (acid pickling, anodizing, etc.). Aluminum foam can be anodized like aluminum, but removal of the electrolyte from the structure is required
to avoid corrosion. Owing to the microporosity of the surface skin, painting with
viscous paints is not problematic. Baking of the paint can result in the formation of
bubbles (blisters in the surface paint) caused by the expansion of gases in pores
and their attempts to escape through surface microcracks. For this reason it is recommended to leave some bare areas to enable the foam to breathe on temperature changes. If the foam has to be sealed completely, the coating process should
be performed in vacuum until the coating solidifies.
Joining
For practical applications of cellular metals, joining techniques are required. As the
structure of cellular metals is similar to that of wood, methods used for joining
wooden parts, especially gluing and bolting, can also be used with cellular metals.
Moreover, methods of joining metal parts, such as welding and soldering, are also
possible.
Joining cellular metal parts with bolts is simple, quick, flexible, and cost effective: the joint is also detachable. The strength of the joint depends primarily on
the density and pore size of the foam and increases if foam with a surface skin
is used. Higher strength is achieved with foams having finer porosity (finer porosity provides a larger contact surface for the thread of the bolt). The strength of the
joint can be significantly improved if the drilled hole is filled with a suitable adhesive.
Brazing and Soldering
The brazing or soldering of metallic foams is still questionable. The metal surface
is always coated with various oxides, depending on the alloy used. Oxides impede
the formation of brazed or soldered joints and therefore they must be removed before or during the soldering. They can be removed either mechanically (scraping)
or chemically (using fluxes). Scraping of oxides usually takes place under a layer of
melted solder to avoid subsequent oxidation. However, this is a laborious operation
demanding a skilled worker and it is costly. The strength of the soldered joint is
usually lower than the tensile strength of the foamed sample owing to the negative
effect of solder on the strength of the basic alloy near the joint. Flux can be used
only for foams with a compact surface skin, otherwise there is a danger of it penetrating into the porous structure. It cannot be fully removed from the pores, which
increases the danger of subsequent corrosion. As small cracks always appear on
the foam's surface, soldering with premature fluxing can be used for sandwich
structures. Foams without a surface skin can be soldered without fluxing, but
the strength of the joint is substantially lower than for samples with the surface
skin (oxides cannot be fully removed from the surface of open pores by scraping).
Glued Joints
Metallic foams can be glued with all types of adhesive recommended for gluing the
cell-wall metal. These joints can be quickly and easily made and their strength
often exceeds the strength of the basic foam structure. Because of the simplicity
and flexibility, glued joints can be recommended for preparation of permanent
joints.
73
74
However, glued joints also have disadvantages: low thermal stability, different
coefficients of thermal expansion for the adhesive and the bonded material, decreasing strength of the glued joint due to aging of the adhesive, and formation
of an insulating layer (thermal and electrical) are some of them. Moreover, if
foams without a surface skin are glued together, a greater consumption of adhesive
(weight increase, cost) has to be expected because of the significantly enlarged surface area.
Welding
Metallic foams can be also welded, if the weldability of the cell-wall material is
good. Welded joints are very promising, but the techniques need further development and the costs are still relatively high for practical application. Welding problems arise mainly from the very low and variable thickness of the surface layer.
This obstacle can be partially overcome using controlled heat input into the
joint, by laser-welding for example. Good joints can be obtained if the foamable
material is used as filler. Welding in combination with such a filler can provide
a method for repair of foamed parts.
Another possibility for joining cellular metals is the use of integral foams containing the parts made from bulk material, which are built into the foamed structure and if possible diffusion bonded to the foam during foaming. These parts provide the contact counterparts of the foam for a joint. The bulk material can be
either locally placed inside the foam if the position of joint is known before foaming (Fig. 3-1a), or it can create a large portion of the foam's surface (Fig. 3-1b). The
material of the built-in element should be compatible with the cell-wall alloy with
respect to corrosion resistance and thermal expansion. Locally distributed elements
need precise positioning and suitable fastening to avoid displacement due to foam
motion during foaming, which makes the foaming procedure more expensive. The
use of various coversheets that are diffusion bonded to the foam during foaming
can eliminate the higher costs and lower flexibility of the built-in elements, on
the other hand the higher overall weight must be taken into account.
a)
b)
Joints made using bulk aluminum parts diffusion bonded to the AlSi12 foam as:
a) locally placed element, b) 3 mm thick Al coversheet.
Figure 3-1.
Further Reading
3.1
Cellular metals are a new class of materials with low densities and novel physical,
mechanical, thermal, electrical, and acoustic properties, which offer potential for
lightweight structures, for energy absorption, and for thermal management [1].
In order for their outstanding material properties to be used, they have to be machined and processed to build structural components. The relevant techniques are
described in this section.
3.1.1
High-Temperature Forming
At present cellular metals produced from the melt, by deposition, and by solid-state
processes are available. They have very valuable properties in terms of density, stiffness, and energy absorption, but their potential usefulness is limited by the difficulty of forming or fabricating finished parts. A forming process is necessary for
the manufacture of 3D structural parts and a joining technique is required to fabricate such parts into composite components.
75
76
Foams produced from melts, in contrast, are much less expensive but there are
problems with the forming properties of foamed cellular metals. At ambient temperature, forming the foams gives rise to fractures or to the collapse of the pore
structure in the deformed zone. Local stress and strain concentrations can occur
causing failure at the weakest parts of the cellular structure [2].
Forming at elevated temperatures facilitates the shaping of metallic foams without failure and with the pore structure being retained. At high temperatures the
yield stress and the deformation limits can be significantly reduced so that yielding
can occur. Local stress concentrations can be reduced by creep relaxation. The details of the process are described in the following section.
Figure 3.1-1.
Figure 3.1-2.
Left) working area in Al Si phase diagram [3]. Right) workpiece formed in a steel
mold [3].
The workpiece in Fig. 3.1-2 was formed using a steel mold in a chamber furnace
in the temperature range of the solidus. The forming force in this case was gravity.
The mold was reproduced well, including the corrugations, but a disadvantageous
temperature management and locally higher stresses at the beginning caused a collapse in contact areas at the edges.
Figure 3.1-3. TiH2 (left) and gas injection foamed (right) foams: influence of temperature and
foaming agent on cell structure.
77
78
weight of the bending wing. As can be seen the material can be formed without
collapse of the cell structure, and relatively large bending angles are possible.
Owing to the properties mentioned above, the temperature range and with it the
working area and the process window can be expanded. Thus, this material is very
well suited to high-temperature forming process in contrast to the TiH2 foamed
materials, which only show a small process window.
3.1.2
Machining
Conventional cutting and machining techniques like sawing, milling, or drilling all
are suitable for processing metal foams. But using conventional machine tools
causes surface distortion or damage to low-density metal foams [5]. Nevertheless,
the conventional methods are satisfactory for most applications in which metal
foams have to be machined [6, 7].
For accurate cutting electrodischarge machining (EDM), chemical milling, waterjet cutting, high speed cutting, or diamond sawing are recommended [5]. This is
especially necessary for machining metallic foam specimens for testing, to avoid
cell damage that would influence the properties of the samples.
Owing to the cellular structure of metal foams there are different methods of
machining the material. The appropriate technique should be selected according
to the required quality of the cutting surface. For cutting of metallic foams reinforced with hard particles like SiC, a diamond cutting tool might be required.
3.1.3
Coating
Owing to their properties, metal foams have great potential in lightweight construction. At present there are few applications for metallic foams because
foams with open surfaces show relatively low mechanical strength and are difficult
to join to other parts. So metallic foams are not yet well suited to structural applications [8]. To further enhance the stiffness and strength of the foams they are
often used as composites with roll-plated or adhesive bonded cover sheets. The
cut surface of metal foam displays open cells and a rough texture, and is vulnerable
to local damage [5]. It can be filled with an epoxy or other resin, or, in the case of
syntactic structures that have a natural skin, the surface can be coated by conventional methods. These manufacturing methods are mostly limited to plane surfaces and do not allow complex shaped parts. By using thermal spraying techniques almost any desired form of foam-spray-deposit composites can be produced.
Furthermore, there are considerable advantages concerning recycling and usage at
elevated temperatures compared with adhesive-bonded composites [8].
79
80
treated to roughen and to activate it. This is generally done by corundum blasting
[8].
Currently available foamed semi-finished products made via the powder-metallurgical route have a thin smooth surface skin. This surface is often damaged during the blasting process, which opens the underlying pores. Thus, a surface treatment that is more suitable for foams is required because the spraying process is
not able to fill the open cells and form a new plane surface. A suitable approach
may be microblasting, in which the semi-finished foam products are blasted
with low pressure and with particles less than 100 mm in diameter. Although the
process parameters have not been fully investigated yet, the first results of the microblasting technique have been very satisfying.
Figure 3.1-6.
81
82
ses and leads to a relatively small reduction of the hardness of the spray deposit.
One disadvantage is the high density of NiAl5 compared with aluminum alloys.
The use of highly tensile solidifiable aluminum alloys as spray materials is the subject of investigations at the present time [8].
Another technique more complicated than wire arc spraying but producing
much better spray deposits is plasma spraying (Fig. 3.1-7). The spraying material
is injected into an inert gas plasma as a powder and then heated and accelerated in
the direction of the substrate. The material is an AlSi30 powder, which in conjunction with suitable process parameters leads to relatively hard deposits when
sprayed. A practical thermal treatment of these spray deposits and their further mechanical properties are currently being investigated.
Thermal spraying is a suitable and flexible method of producing composites with
metallic foam cores. But investigations are still necessary with regard to the enhancement of the mechanical properties of the deposits as well as the pretreatment
of the metallic foams.
Figure 3.1-7.
Left) spray deposit on aluminum foam [8]. Right) plasma sprayed deposit [8].
References
3.2
Introduction
This section summarizes conventional joining techniques that are feasible for compound structures utilizing aluminum foams.
83
84
which have the advantage that no additional hole has to be drilled into the sheet
material.
The performance of the mechanical joints can be improved if the fastening elements are additionally smeared with epoxy adhesives before driving them into the
foam [4,5].
A further possibility for joining foams with mechanical fasteners is the application of integral foams [3], where the foam contains parts made from bulk material.
These parts, screw sockets for example, are integrated into the foamed structure
and bonded to it during the foaming process. However, the locally distributed elements need precise positioning and suitable fastening in order to avoid a displacement caused by the foam motion during foaming. Consequently, not only is the
weight of the cellular structure increased by integrating these parts, but the cost
of production as well.
Gluing
Gluing has been described as a feasible joining technology for aluminum foams
[6]. Different gluing systems have to be distinguished. One-component systems
are, for instance, epoxy-based and cure for approximately 30 min at 180 hC. The
process conditions of these adhesives make them very interesting for car manufacturers, as the glue can cure during cathodic electrocoating and no further working
steps are necessary. Epoxy-based systems are available as conventional glues or as
expanding glues. The latter ones are very convenient for bridging high tolerances
and for integrating foam bodies into hollow extruded profiles. In case a thermal
curing process is undesired, two-component systems are available that cure at
room temperature. By combining riveting and gluing it is easily possible to fix
the foam body in the structure while the curing of the glue takes place. Additionally, the properties of the composite are improved in certain loading cases.
3.2.2.2
Welding
Welding processes that have been shown suitable for joining aluminum foams are
laser-beam welding [6] and ultrasonic (US) welding [7].
With the latter process, various geometries of the joining area are possible. They
reach from quadratic joints with a maximum area of 100 mm2 to roller seam welds
with a width of 1 5 mm and a maximum length of 500 mm. Finally circular bonding areas with typical diameters of 18 mm are producible. A characteristic of the
US welding process is that it is a joining process in the solid state based on friction
and consequently not a real welding process, where both joining partners are transferred into the liquid state for joining [7].
Laser-beam welding has been shown as a viable joining technology for aluminum foams [6]. Owing to its characteristics such as the keyhole effect and the
minimal heat input, deeper parts of the structure can be integrated into the
joint as well. Fig. 3.2-1 shows the principle mechanism of surface welding and
the key-hole welding effect for bulk and foam material.
3.2.2.3
Figure 3.2-1.
Soldering
Various Sn- and Zn-based solders can be used for fabricating structures using Al
foams [3]. Reliable joints are said to be possible without the use of fluxes, just
by scratching the oxide layer.
Another suitable technology is the S-Bond process, developed by Euromat [8].
This solder has a chemical composition of SnAg4Ti4 (wt.-%) and a melting
range of 220 229 hC. The S-Bond process is a flux-free soldering process that is
characterized by the mechanical ripping of the oxide layer that has formed around
the molten soldering material. The oxides of the substrate layer are partially ripped
during the soldering process, so that a metallurgical interaction of the soldering
material and the substrate can take place. Additionally, the soldering material
creeps below the oxide layer of the substrate and wets its surface totally. S-Bond
is available as wire, rod or foil, which makes this process an interesting alternative
for extensive joints.
Though flux-free soldering technology can be transferred to aluminum foams,
corrosion preventing measures have to be taken as the Sn- and Zn-based alloys
can cause corrosion problems in combination with aluminum.
85
86
Brazing
Brazing (Ti450 hC) is a feasible joining process for aluminum foams [9]. Though
noncorrosive fluxes are available for brazing, it is advisable to coat the structure to
protect it from humidity. The brazes are preferably Al-based and the furnace atmosphere should be inert.
An alternative is the use of a porous filler material [9], which expands during the
brazing process (Fig. 3.2-2). Thus a sudden change in porosity from foam to bulk
material is prevented. The working temperature of the filler material has to be low
enough that the porous base material does not deform during the expansion of the
filler material.
As outlined, there is a variety of suitable joining technologies available for fabricating structures utilizing aluminum foams. However, for a practical investigation
the number of technologies has to be limited to one or two processes per category.
Table 3.2-1 gives an overview and some characteristic properties of relevant joining
technologies.
Figure 3.2-2.
Joining of Al foam and solid Al by brazing with a filler that expands to a foam [9].
Joining
technology
Formation of
the joint
Joining
temperature
Joining class
Detachability
Riveting
point-contact
RT
mechanical
Screwing
point-contact
RT
mechanical
Gluing
area-contact
RT to 180 hC
adhesive
Riveting/Gluing
area/pointcontact
RT to 180 hC
mech./
adhesive
Flux-free soldering
area-contact
250 hC
thermal
Laser-beam welding
line-contact
Melting Point
thermal
3.2.3
Table 3.2-2.
T-joint
Overlap-joint
Blind riveting
Screwing
Gluing
Riveting/gluing
Soldering
Welding
with restrictions
with restrictions
3.2.4
The following text concentrates on processes for producing foam sheet composites [2,11]. Though there are other processes, such as the roll-bonding process
[1], further technologies have to be investigated, for integrating foam bodies into
hollow extruded profiles or for the partial stiffening of structural parts. In most
87
88
cases, foam sheet composites offer only one-sided accessibility. This has to be
taken into consideration when selecting a feasible joining technology. Under
these circumstances, all joining technologies listed in Table 3.2-1 are feasible.
Microstructural Investigations
Figure 3.2-3 shows micrographs of the investigated foam sheet joints.
3.2.4.1
Figure 3.2-3.
a) Screwing
b) Riveting
c) Gluing
d) Soldering
e) Laser-beam welding
Figure 3.2-3.
89
90
or the substrate. The bonding of the glue and the surface might either be caused by
ionic bonding or polar interactions [12]. Gluing is a feasible joining technology for
foamed metals, independent of the presence of a casting skin. Though the area in
direct contact with the sheet material is drastically reduced in the absence of a casting skin, the viscosity of the glue is low enough to fill the pores. The loss of surface
can be compensated for by an additional form-fitting joining mechanism.
Soldering (Figure 3.2-3d)
Like gluing, soldering can be used as a joining technology generating an area-contact joining mechanism. In spite of the extensive oxide layer on the metal foam,
interdiffusion zones form between solder and substrate materials, when applying
S-bond. For soldered samples, the presence of a casting skin is advisable. As already described, relative movement between the parts is necessary during joining
with a flux-free solder like S-bond [8]. Consequently, both surfaces have to be as
plane as possible, so that the wetting of the substrate can take place everywhere.
If there is no casting skin, the solder flows into the pores, but in contrast to
glued samples does not form a material-fitting joint with the cell walls, as the
oxide layer of the molten solder in the pores cannot be torn properly during the
relative movement of the parts. Consequently, the remaining joining area would
be limited to the number of the present cell edges, which would result in a reduced
strength of the structure.
Laser-Beam Welding (Figure 3.2-3e)
The connection of the sheet to the foam with a laser is based on two principal joining mechanisms: a material one and a shape-related one. The heat input of the
laser causes the melting of the sheet material and the underlying foam structure.
After resolidification a metallic connection is created. However, this connection
exists only between the molten sheet material and the molten material of several
cell structures located close to the interface. The keyhole effect supports this mechanism. Because of the deep penetration of the laser beam, additional cells that
are located in close proximity to the sheet surface can be included in the joining
zone. Additionally, the concentrated heat input of the laser beam causes only minimal melting of foam structures outside the original joining zone. On the other
hand, the molten material from the cover sheet and filler material flows into the
pores of the foam structure during the welding process and fills them partially.
The additional material prevents the formation of a seam groove and is responsible
for the shape related connection mechanism.
Figure 3.2-4.
Table 3.2-3.
Testing speed
Distance of the
bending die/stamp
240 q 25 q 10
5 mm/min
200 mm / 100 mm
145 q 25 q 10
5 mm/min
free distance: 45 mm
145 q 25 q 10
5.7 km/h
105 mm
91
92
Pull-Out Tests
Pull-out tests were carried out with structures utilizing mechanical fastening elements. After preliminary investigations with a selection of conventional fastening
elements, all fasteners were excluded that form the joint according to a frictionalfitting joining mechanism. Though good results with frictional-fitting fastening
elements could be measured with large volume samples [5], the thickness of
foams for technical applications such as the partial stiffening of structures should
be limited to less than 10 mm. In this case however, there is not enough material
left for the frictional interaction between cell edges and the fastening element, so
that it will be torn out at low loads. Figure 3.2-5 shows some characteristic
load deflection curves of joints utilizing mechanical fasteners with an additional
form-fitting joining mechanism.
Riveted samples show a larger range of deformation as the bulge has to be pulled
through a hole of smaller diameter. This means that a great amount of deformation work has to be done. In contrast to this, the diameter of the investigated
screws reduces to the tip of the screw. As a consequence, not much deformation
work can be done when pulling out the screw, after the maximum load has
been reached.
Obviously, the pull-out load is dependent on the density of the foam material.
The effect of relative density on the maximum pull-out load, Fr, is nonlinear
with a power of about 1.5 [5]
Fr Ar3=2
Figure 3.2-5.
loading.
(1)
Qualitative load deflection curves for screwed and riveted samples in tensile-shear
Tensile-Shear Tests
Tensile-shear tests were carried out for the joining technologies listed in Table 3.2-1.
The maximum shear loads for the joining technologies are given in Table 3.2-4.
There are two different failure mechanisms that apply to point-contact joining
technologies or area/line-contact joining technologies. Laser-beam welded and
Joining method
Riveting
Screwing
1282 e 130 N
Riveting/gluing
1548 e 182 N
Gluing
2043 e 220 N
Flux-free soldering
2112 e 596 N
Laser-beam welding
580 e 116 N
glued samples tend to fail in the foam close to the joining zone. Consequently, the
mechanical properties of the cellular structure influence the behavior under tensile
shear loading. This is observable at a maximum load that represents the mechanical properties of the foam. After the maximum load is passed, the foam tears almost brittle. Laser-beam welded samples show a lower maximum load than glued
samples as their joining area is smaller. Consequently the failure load is highly
dependent on the number of welding seams.
In contrast, riveted and screwed samples show ductile failure behavior under
tensile-shear loading. The fastening element compresses parts of the cellular structure during the deformation. When the remaining cross section of the foam body
is too small, the foam breaks, which is shown by an abrupt decrease of the load.
Influence of the Foam Density on the Joint Strength for Line-Contact Joints
Figure 3.2-6 shows that the strength of welded samples is highly dependent on the
relative density of the foam [2], i. e. the number of the cell walls in the area of the
joint. In these experiments the casting skin was removed before joining so that the
resulting joining mechanism is based on a fusion of the sheet with the cell structures as well as by form-fitting due to a flow of the melt into the porous structure.
The failure of the joint is observed to start close to the welding seam.
The failure behavior can be described by a correlation between the foam density
and the strength of the joint.
s Metalfoam
Cr2
max
(2)
rmin S ar
s Metalfoam
max
(3)
There are two ways of assessing the dependence of the strength of laser-beam
welded composites on the foam density. For example, the strength of massive
sheet sheet joints of about 210 N/mm2 (Fig. 3.2-6) would result in a relationship
according to a power law (Eq. 2). In this study however, the joining partner is not a
massive material but a powder-metallurgic produced aluminum. As a consequence, the values of the massive material cannot be reached with a density of
93
94
a)
b)
Figure 3.2-6.
Influence of the density of the Al foam on the strength of laser-beam welded joints.
2.7 g/cm3. The values of composites utilizing a metal foam with a casting skin indicate a linear relationship between the foam density and the strength of the joint.
Here, the casting skin must be assessed as a massive joining partner and the
resulting density of the casting skin is 1.8 g/cm3 (linear regression of Eq. 3).
The solution of the power law equation gives a lower value of 1.1 g/cm3.
Bend Tests
Influence of the Joining Technology on the Properties in Four-Point Bending
With the help of the slopes in the linear section of the load deflection curves in
four-point bending, it is possible to calculate the effective bending stiffness
(EJ)eff of a sandwich element according DIN 53 293. The optimum joint is an
area joint, so glued sandwich structures show the highest effective bending stiffness, which is equal to the theoretical bending stiffness (EJ)theo. Further results
show that a decrease of the joining area by utilizing other joining techniques
leads to a reduction of the stiffness of the structure.
Therefore, riveted samples show the worst properties due to the puncture joint.
The stiffness of the structure is additionally reduced by the holes that are necessary
Figure 3.2-7.
for integrating the rivets. Welded and riveted/glued samples show an average stiffness, like soldered specimens.
As the integral of the load deflection curve (equal to the energy absorbed during the deformation) increases with rising stiffness, the stiffness of the composites is mirrored in the values of energy absorption up to a deflection of 10 mm
(Fig. 3.2-7).
Related to their mass, structures utilizing a steel sheet (applied joining technology: gluing) show the worst properties owing to the high weight of the sheet
material. Composites utilizing a tissue on the tensile side of the structure are
the lightest, but they have a low stiffness owing to the tissue. In consequence,
the amount of energy that can be absorbed is comparatively low.
Four-point bend tests with a constant bending moment along the sample serve
for evaluating the stiffness of a sample. In the case of a three-point bend test
the maximum bending moment is in the area of the bending stamp. Thus,
three-point bend tests serve for evaluating the strength of the structure.
Influence of the Joining Technology on the Properties in Three-Point Bending
As in four-point bending, the joining technology has a strong influence on the
properties of the composite. In contrast to the four-point bend tests, glued/riveted
samples show the best properties in this loading case. Structures that were simply
glued failed owing to cracks in the foam core. By combining the processes, the rivet
relieves the foam core from high stresses and makes the structure less dependent
on inhomogeneities of the foam. This is confirmed by the comparatively low variance of the results. Apart from the structures utilizing a tissue, all samples vary
around the same value for the maximum load. For realizing a good energy absorbing behavior a ductile deformation behavior is necessary as well as a high maximum load. According to Fig. 3.2-8, which shows the amount of energy absorbed
during three-point bending after a deformation of 10 mm against the mass of
the structure, glued, riveted/glued, soldered, and laser-beam welded structures
95
96
Figure 3.2-8.
meet this requirement. Though the strength of the tissues is high enough, it fails
after a comparatively low deformation owing to catastrophic fracture.
Displaying the amount of absorbed energy in Fig. 3.2-7 to 3.2-10 shows that all key
figures located on the same straight line through the origin of the diagram have
the same mass specific value. With an increasing slope of the line, the mass-related
properties increase as well. The goal of optimization is therefore to increase the
slope of the line on which the values are located.
This cannot only be achieved by a variation of the joining technology, but also by
the density of the foam core.
Influence of the Density of the Foam Core on the Properties in Three-Point Bending
The results of three-point bend tests with structures utilizing a foam body without
casting skin and an average density from 0.3 to 1.1 g/cm3 show that there is an almost
linear dependence of the maximum load on the density of the foam core in the investigated density range.
Figure 3.2-9. Influence of the weight of the sandwich structure on the value of the absorbed
energy up to a deflection of 10 mm (see Table 3.2-3)
Comparing the performance of sandwich panels where the casting skin was still
on the core material to those without casting skin reveals that the sandwich panels
with core material without casting skin have a higher specific energy absorption
(Fig. 3.2-9).
As the foam is compressed in the area of the bending stamps during three-point
bending, the loading mode in this area is consequently similar to a compression
test. However, the casting skin is orientated perpendicular to the load and cannot
enhance the energy absorbing properties. Thus the casting skin just decreases the
mass-related energy absorbing properties in bending.
Influence of the Joining Technology on the Properties in Dynamic Three-Point
Bending
Dynamic three-point bending of sandwich structures serves for simulating sudden
impact, like in crash situations. The goal is to absorb an impact energy that is as
high as possible. This can only occur if the sheets stay bonded well to the foam
core during the whole deformation. Figure 3.2-10 shows the influence of the
different joining technologies on the impact energy that can be absorbed
against the mass of the structure. The results of the dynamic three-point
bend tests basically confirm the results of the quasi-static tests. Glued and
glued/riveted samples show the best properties, followed by soldered specimens, owing to the ductile deformation behavior of the joint in bending.
Welded samples show a high variance. The high dependence of this joining
technology on the porosity of the foam increases with the testing speed. If
some parts of the structure are not joined well, owing to an uneven surface
of the foam or because of big pores in the area of the welding seam, and
if these areas are orientated unfavorably during the deformation, the composite
structure fails almost without absorbing any energy. Under optimum conditions
however, most of the kinetic energy can be absorbed. In this case, continuous
seams show better results than step-welded seams. Step-welded seams are
usually used in order to stop cracks propagating through the seam. However, in
Figure 3.2-10. Influence of the joining technology on the energy absorbing behavior of a sandwich structure in dynamic three-point bending (see Table 3.2-3).
97
98
The basic fields of application of aluminum foams are energy absorbing structures
and stiffening of structural elements. The following sections show the transferability of the described joining technologies to structural elements.
Tubes
The integration of foam bodies in tubes for energy absorbing structures is well described elsewhere [13,14]. The results show that gluing as well as laser-beam welding are feasible joining technologies for this application. Figure 3.2-11 shows an
X-ray of a crash-absorbing element partially filled with Al foam and a detail of
the joining area.
Regarding the triggering of the sequential buckling process, laser-beam welding
shows some synergetic effects in axial loading, as shown previously [13,15]. The
initial peak load of the characteristic load deflection curve in axial loading can
be reduced, which leads to an increase in the effectiveness of the composite structure in axial loading.
3.2.5.1
Figure 3.2-11. Foam body integrated in an Al tube by laser-beam welding for an application as a
crash element in the front side member of a vehicle [13].
3.2.5.2
Hat-Profiles
Figure 3.2-12. Integration of a foam body into a hat profile for improving the torsion, bending,
and energy absorbing behavior of the structure.
Figure 3.2-13. Influence of the joining strategy (one- or three-sided) and the joining technology
on the deformation behavior of a hat-profile in dynamic three-point bending.
99
100
Owing to the optimum contact area of glued samples, the sheet material stays
in contact with the foam during the whole deformation of the composite structure. In the welded sample, however, the sheet can deform independently of the
foam in large areas, which reduces the energy-absorbing potential of the structure in a defined interval of deformation. Independently of the joining technology, three-sided joining improves the energy-absorbing properties of the structure.
Torsion Tests
The torsion tests shown in Fig. 3.2-14 show that glued samples have better properties than welded ones. Fewer folds form during the deformation process, which
means that more deformation energy can be absorbed by the porous structure.
In laser-beam welded samples, the welding seams that fix the foam core are the
weakest part of the structure and tear during the loading process.
Corresponding to the bend tests, the experiments with foam-filled hat profiles
show that with increasing area of the joint, the properties of the composites increase.
Figure 3.2-14. Influence of the joining strategy (one- or three-sided) and the joining technology
on the deformation behavior of a hat-profile in torsion.
3.2.6
Summary
In a final overview, Table 3.2-5 and Table 3.2-6 show the properties of the investigated joining technologies for different cases of loading, including a technological
assessment of the joining technologies.
Tensile
shear test
Four-point
bend test
Three-point
bend test
Dynamic
bend test
Riveting
++
Screwing
Gluing
++
Riveting/
Gluing
++
++
++
Laser beam
Welding
Table 3.2-6. Final evaluation of the investigated joining technologies (++ very good; + good;
o overage; bad)
Accessibility
Costs
Recycling
Functionality
Freedom
of design
Transferability
on structural
parts
Riveting
++
++
++
Screwing
++
++
++
Gluing
++
++
Riveting/
gluing
++
Soldering
++
Laser-beam
welding
++
++
101
102
References
3.3
Encasing by Casting
Foam parts prepared by expansion in a mold show a closed surface skin with a
thickness comparable to the cell-wall thickness of about 200 mm. Even in cases
where the surface near cells are compressed yielding a few layers of cell walls
along the outer skin, it is not suitable for carrying high loads, as required for structural applications. Casting aluminum around a foam can create components where
a low density foam core is completely surrounded by a massive outer shell. The
shell can be designed in such a way that additional functions besides its load carrying ability can be fulfilled. As only one processing step is required to produce such
a foam core component, production is expected to be very economical.
Potential applications for this kind of foam core castings are space frame nodes,
knuckles, control arms, cross-members, and stiffness providing structural components. The described method also offers the potential to produce quasi-hollow castings where conventional salt or sand cores are replaced by permanent foam cores
in complex geometries.
The various casting methods for the massive outer shell differ by the acting pressures, the die-filling time, and the solidification time, which is mainly determined
by the temperature of the die. Generally, casting of the shell is not completely
straightforward due to the limited compressive strength of the foam cores and
the very thin surface skin, which is not without flaws [1]. A further difficulty results
from the fact that methods have to be developed for fixing the cores.
The following text is based on three publications: squeeze-casting, Kretz and
Kaufmann [2]: high-pressure die-casting, Heinrich, Korner, and Singer [3,4]; and
gravity die-casting, Simancik and Schoerghuber [5].
3.3.1
103
104
tion-molding technique is thicker than that of the furnace baking method. This is
an advantage for encasing but it also increases the weight of the core. Figure 3.3-1
shows several core geometries used for different encasing techniques.
Core Attachment
A suitable core attachment system has to perform two functions: first it has to keep
the core in place in the die when the die is open and during the movement of die
closing. The more important second function is to maintain the desired distance
between the foam core and the die wall during the casting process because this distance determines the wall thickness of the castings.
One method to realize a suitable attachment of the core is by creating elongated
wedge-shaped spacers during the foaming process of the cores (Fig. 3.3-2).
These spacers have to be long enough to transfer the acting forces to the foam
core without damage. It is very important to place the spacers corresponding to the
melt flow to avoid flow barriers resulting in dead zones. This is especially true for
die- and squeeze casting, where the melt freezes quickly.
3.3.1.2
Mechanical Properties
The aluminum alloys used for the foam cores are wrought or casting alloys. Generally, the porosity of the surface, from the appearance of cracks or open cells, is
higher for the casting alloys. On the other hand, due to their smaller cell size, casting alloys are more tolerable against penetration of melt. This is especially true for
small samples.
Besides defects already present there are two important properties of the cores
that determine the processing parameters during high-pressure casting: the
foam compression strength (global strength s CG) and the compression strength
of the surface skin (local strength s CL). The global strength s CG (Fig. 3.3-3) defines
the maximum allowed pressure in the process without foam compression [3].
The samples prepared of the wrought alloy (Fig. 3.3-3) show a significantly
higher scatter of the values compared to the casting alloy. This is a result of the
small sample thickness of 10.5 mm and the coarser pore structure of the wrought
alloy foams in the present case.
It is observed that infiltration of foam cores already occurs at pressures that are
much lower than s CG. This is either the result of open pores in the surface or the
product of a local fracture of the surface skin. As a consequence, an indefinite pore
volume is infiltrated. The local compressive strength s CL can be measured by a
compression test where a liquid (water) exerts a homogenous pressure on a
foam surface of 25 mm q 25 mm (Fig. 3.3-4) [4].
From simple mechanical considerations one finds that the maximum bending
stresses acting in the cell walls that make up the surface are proportional to
3.3.1.3
p(D2/d2) t p(1/d2r2)
where p pressure, D cell diameter, d cell-wall thickness, r foam density.
As a result, the local compressive strength s CL increases with increasing density.
Figure 3.3-3. Global compressive strength s CG versus density of several Al foam samples from
plates produced via the furnace-baking method [4].
105
106
Figure 3.3-4. Average values of the local compressive strength versus density of several foam
samples produced via the furnace-baking method [4].
There is a strong scatter of strength values due to the inhomogeneous cell structure of the foams.
The results in Fig. 3.3-4 demonstrate that the strength of the foam surface skin
has to be improved for squeeze and die casting, which are characterized by a very
high pressure. For gravity die casting the global strength as well as the local
strength are not so important.
a)
Figure 3.3-5.
b)
c)
3.3.2
107
108
Figure 3.3-6.
cores are not infiltrated and not deformed. However, the processing window for
achieving good samples without core infiltration or core deformation is a very narrow one.
Die Casting [3] [4]
Essential for die casting is controlling dwell pressure with the help of a real-time
control system [3,4]. In the first phase of the conventional die-casting process the
die is filled with melt as fast as possible. In the second phase very high dwell pressures are applied to balance the shrinkage and to minimize porosity of the castings
by compressing the entrapped gases. Because of the high velocity of the castingpiston a high pressure peak appears at the end of the first phase when the die
is completely filled resulting in massive foam infiltration and collapse (Fig. 3.3-9).
In order to avoid this peak, a real-time controlled die-casting machine is indispensable. With real-time control it is possible to control the casting velocity and
the casting pressure at every step of the casting process. That is, the piston can
be decelerated at the end of the filling phase to prevent the pressure peak.
To use a real-time controlled machine an accurate dosage system is required. The
castings were performed with the standard casting alloy AlSi9Cu3. The melt temperature was about 720 740 hC, die temperature about 250 270 hC. Casting investigations were carried out on two demonstration parts (Fig. 3.3-10).
Figure 3.3-9. Comparison of dwell pressure with and without real-time controlled
die-casting machine [4].
109
110
Figure 3.3-10. Two different demonstration parts with foam core inserts encased by die casting:
a) trapezoid, b) plate [4].
As core materials the casting alloy AlSi10 and the wrought alloy AlMg1 were
used. The upper bound of the pressure inside the die is limited by the global compression strength of the foam core. It is about 60 100 bar and can only be raised
by increasing the foam density.
a)
b)
Casting experiments with coated and uncoated foam cores were performed for a
plate-like geometry with different specific casting pressures pNS. The specific casting pressure is the operating dwell pressure in the casting chamber. It has to be
taken into account that the actual acting pressure inside the die is much lower
because of friction. Figure 3.3-11 shows the infiltrated pore volume VI and the
porosity in the skin versus pNS.
As expected infiltration increases with increasing pNS whereas porosity in the
skin decreases. By using coated foam cores a significant reduction of core infiltration is possible. In accordance with the tests to determine the local compression
strength, the plasma-sprayed plates show the best results. More recent results
achieved with optimized die design and process conditions (evacuation of the
die) demonstrate that the porosity of the skin can be reduced even to less than
5 % without infiltration of the core. The minimum skin thickness is about 2 mm
for die geometries evaluated so far.
Gravity Casting [5]
The foam core samples (AlSi12 and AlMg1Si0.6) used had a density of
0.8 0.9 g/cm3 and were prepared by injection of the foam into sand molds
(Fig. 3.3-12) using the injection-molding technique. They were inserted into
an existing gravity casting die (steel) with conventional gating and risering.
The casting alloy AlSi9Cu3 at 740 hC was used to cast the outer shell around
the foam cores.
In contrast to die- or squeeze casting the acting pressures during gravity casting
are insignificant and much smaller than the local compressive strength of the foam
cores. That is, as long as there are no large cracks in the foam surface infiltration or
compression due to high pressures is not observed during gravity casting. No melting of the foam core could be detected in the experiments performed (Fig. 3.3-13).
Pouring liquid metal around the foam core results in heating up and consequent
expansion of the gas inside the pores. In contrast to the high-pressure casting
methods the gas is not prevented from diffusing from the core into the melt
where it forms bubbles. That is, the expanding gas can cause porosity in the
shell or even destroy the integrity of the casting because of gas bubble formation.
This expansion of the gas and the deterioration of the shell can be minimized by
preheating the cores before insertion.
Figure 3.3-12. Pressure-less casting: Foam core, encased core, and cross section with foam core
(height of the foamed part 25 cm, maximum width of the cross section 12 cm) [5].
111
112
Figure 3.3-13. Aluminum casting with different foam cores: left, AlSi12 foam; right, AlMg1Si0.6
foam (diameter of the foam filled tubes 6 cm) [5].
Another advantage of preheating the cores is that cold shuts, which appear in
regions of the castings with the lowest thickness of the shell, can be avoided.
The preheating of the cores leads to a lower heat flow from the melt into the
core, thus enabling reduction of the wall thickness in comparison to usual sand
cores, which can not be preheated. With preheating temperatures of 400 hC the
minimum shell thickness is found to be about 1.5 mm.
Mechanical bonding by flow of liquid metal into the outer foam structure supported by intentional weakening of the surface skin, e. g., by sand-blasting. Disadvantage are that the weight of the casting increases and the bonding occurs
only locally and is difficult to control.
Metallurgical bonding by coating the cores with various agents supporting diffusion through the aluminum oxide layer. With a suitable metallic coating a metallurgical bonding can be achieved. On this point, however, further research is necessary.
The shrinkage of the cast shell during solidification leads to a pressure fit of the
inserted core. For most applications metallurgical bonding will not be necessary
to achieve the required component properties such as improved energy absorption
behavior or improved damping properties. The absence of bonding may sometimes
even improve structural damping of the part due to additional energy dissipation at
the shell/core interface [5].
References
3.4
Sandwich Panels
For structural applications metal foams are often used in combination with conventional dense metal structures such as sheets, columns, or more complex-shaped
hollow metal structures. This allows for optimized mechanical properties in a
given loading situation [1]. It also facilitates hiding the metal foam inside a closed
and dense structure, which again is advantageous for corrosion protection. Such
composites containing aluminum foam may be manufactured in various ways.
The most obvious and straightforward one is achieved by adhesive bonding of
pre-fabricated aluminum foams and flat face sheets. However, this approach has
certain disadvantages and is not feasible in all cases. An alternative and preferable
method consists in establishing composites during the foaming process. Foamfilled columns can be produced by inserting foamable precursor material into a column and then heating. The foam will eventually rise and fill the column. Another
possibility is to start with a foamed part and to coat it with aluminum by thermal
spraying [2] thus establishing a dense outer skin. Yet another way is to use the
foam as a core for die-casting [3,4]. Aluminum foam sandwich (AFS) panels
may be manufactured very elegantly by roll-cladding face sheets to a sheet of foamable precursor material, then creating the desired shape in an optional working
step, and by finally foaming the entire composite [5,6] (Fig. 3.4-1). Foaming will
create a highly porous core structure without melting the face sheets if the melting
points of the foam and the face sheets are different and process parameters are
chosen appropriately.
113
114
Process steps for making sandwich panels with aluminum foam cores.
3.4.1
shows no structure in the low magnification chosen. The interface of foam and
face sheet lies on a straight line and is well defined.
Finally, the foamed sample on the right hand side of Fig. 3.4-3 represents an
even later stage of expansion. It exhibits a notably coarser grain size distribution
in the foam and a slightly diffuse boundary between foam and face sheet. The
eutectic phase has grown into the former face sheet material by diffusion processes and has locally amalgamated with the face sheet alloy: one reason for
the excellent bonding between foam cores and face sheets in properly manufactured aluminum foam sandwich parts and the explanation of the absence
Figure 3.4 -3. Metallographic images of sandwich structures with an AlSi7 core. Three different expansion stages are shown: left) unfoamed precursor material; middle) sandwich
115
116
Industrial Application
Figure 3.4 -4. Detroit Show Car 1998 developed by Karmann (Osnabru
ck) for demonstrating
potential uses of aluminum foam sandwich panels. The rear bulkhead and the front firewall (not
directly visible) are made of AFS.
Figure 3.4-5. EUROC 99 concept racing car: left) CAD model of entire car; right) space-frame
structure with AFS parts.
ing parts. The application of AFS parts may also lead to an elimination of noise
attenuation materials because of the low structure-borne sound characteristics of
AFS.
Another area of interest to be investigated in the future is that of exterior panel
closures, e. g., doors, hoods, and decklids. This application depends on the
achievement of a Class-A surface with stamped AFS panels. Taking a hood
stamped out of AFS as an example, there is no longer any need for an inner
panel due to the inherent stiffness of the AFS outer layers, thus reducing expenses for material, tools, and the complete assembly. In spite of higher material
expenses, an AFS bonnet may (depending on the shape) be more cost-effective
than a steel hood up to a production volume of 100 000 units. This is attributed
to the reduction of manufacturing and tooling expenses. This means that light
weight AFS constructions may be applied economically in low and middle production volumes due to the reduced investment costs compared to conventional
steel components.
In general applications, possible additional costs of AFS panels should be compared to the advantages that can be expected. Even if only one of the improved
characteristics of AFS is required for a particular application, the associated increase of expenditure may be of secondary interest. In general, the application of
AFS components will increase overall material costs, so that an enhanced performance must be achieved as justification. In the future, this situation may change
as vehicle operating costs gain more significance due to the higher energy costs. As
a consequence, this implies that the amount of money available for the use of light
materials, which is right now about 2.5 5 Euro per kg weight reduction, will
increase.
Technological Benefits
Besides the advantages already known (the combination of high torsional stiffness
with low weight) further properties of AFS must be taken into account as they also
may have a major impact on the implementation strategy.
3.4.2.1
117
118
Acoustic Properties
Considering the customers comfort requirements, a new lightweight body material is not allowed to show inferior acoustic properties than steel. Therefore, the
good acoustic properties of AFS should be emphasized. Especially cars with aluminum bodies show poor damping properties. To improve this situation, a large
amount of damping material must be added, thus sacrificing the mass saving potential.
A joint research by the TU Dresden and Karmann has revealed that AFS offers a
significantly better acoustic behavior especially in the range 50 400 Hz. Additional
insulation measures may be reduced: weight is saved. The acoustic performance of
special insulation materials will of course not be attained, so that AFS cannot exclusively serve as a sound damping material.
Thermal Properties
Thermal conductivity is another important aspect when selecting body materials.
Due to the entrained air bubbles, the heat transport capabilities of foams are
low. Depending on the density, the thermal conductivity of AFS is reduced to
1/12 1/20 of the conductivity of bulk aluminum. Furthermore, AFS satisfies
most of the fire protection regulations. No adhesives are contained and the AFS
components maintain their shape up to the melting point of 600 hC and in
some cases even above. The exceptional welding characteristics together with minimal distortion justify to characterize AFS as thermally very stable.
Robustness
Sandwich components, as known from aerospace technology, are relatively vulnerable in impact situations. Even small damages of honeycomb panels may lead to a
complete breakdown of the core panel structure. As a result of the metal link between core and panels, this does not occur in an AFS panel. Cracks may only occur
in the core and their expansion is limited. A delamination of core and panels has
not been detected with parts manufactured and tested to date. This is very important since structural body components are not subject to special checks during the
product life cycle
Other Properties
Additional properties of AFS that enhance product performance are: good energy
absorption, recyclability, and low manufacturing time periods for the sandwich components. The foaming process, e. g., takes only 30 45 s even for large parts. Therefore, mass production with a comparably low number of parallel production lines
is possible.
Technical Limitations
When selecting applications for AFS components, their formability and geometry
after the foaming process must be taken into consideration. A constant component
thickness may only be achieved with plane sheets. Complex formed structures will
3.4.2.2
119
120
Process
Details
Laser welding
TIG welding
MIG welding
Pin/bolt welding
by hand, mechanized
Punch riveting
AFS/aluminum; AFS/steel
Riveting nuts/screws
M4
M8
Flow drilling
M4
M8
Riveting
Bonding
Laser Welding
The process of laser welding is suitable for series production. In order to weld face
sheets of 1.2 mm 3 kW power are required. By bluntly welding sandwiches, only
the face sheets are bonded, while the core layers remain unaffected. As shown
in Fig. 3.4-7a the filigree cell structure is not damaged by the local thermal impact.
In welding flat AFS sheets with linear joints, a maximum speed of 10 m/min has
been achieved. If both face sheets have to be bonded, usually the part has to be
turned thus decreasing the welding speed to a rate less than 5 m/min.
3.4.3.1
TIG/MIG Welding
Welding of AFS parts by common welding techniques is also possible. The principle techniques worth mentioning are TIG- and MIG-welding (Fig. 3.4-7b). Both
techniques are exceptionally well suitable both for joining two AFS parts and
also AFS parts with aluminum parts. The advantages of these techniques include
flexible application possibilities, the amount of experience in this field and the low
investment costs. The high degree of stiffness of AFS sandwich parts, their low
thermal conductivity and the resulting low thermal distortion rate minimize the
necessary efforts for clamping and fixing the parts. As in laser welding, only the
face sheets have to be welded. Welding rates for manual TIG welding reach
0.3 m/min, those for MIG welding reach 0.8 m/min. Partly mechanized welding
with a linear carriage may achieve a rate up to 1.3 m/min.
3.4.3.2
a)
b)
c)
Bolt/Pin Welding
Another joining technology of considerable interest for AFS sandwiches is that of
welding bolts (Fig. 3.4-7c). These bolts do not transfer high forces but are employed for fixing cable bundles and wires or as mass contacts. The welding process
is completely controlled and monitored with the help of a welding head with a linear motor. This technology allows for the regulation of variations of the thickness
of the sandwich. As with the other techniques, the core layer remains intact in the
welding area. Even welding a bolt directly onto a flaw, e. g., a void or a large pore
3.4.3.3
121
122
has no impact on the quality of the bonding as the joining zone is obviously limited to the face sheet.
Punch Riveting
Due to the increase of mixed constructions and the problem of thermal impact on
the structure by assembling parts by welding, this joining technology has been
strongly forced in the recent years. Since 1994, Punch Riveting is applied in
construction of the AUDI A8 and has reached its momentary peak in the construction of the A2 with 1800 die casts per car. Figure 3.4-8a shows the joining area as
well as a cut of the fusion. Studies of the settling properties of the core layer have
shown another positive characteristic of AFS sandwiches. If AFS is highly com3.4.3.4
a)
b)
c)
Figure 3.4-8. Different joining techniques for AFS: a) punch riveting, b) riveting nuts and screws,
c) flow drilling.
pressed, its tensile strength drops to a value of 50 % plastic deformation, but regains the original mechanical values of the nondistorted sandwich at maximum
compression. This property is a result of the increasing mechanical clutching of
the collapsed cell structures and the likewise increasing friction. The mechanical
values of compression strength and shear stress do decrease in this range but
they stabilize again at a low level under static as well as under dynamic load.
Flow Drilling
Flow Drilling (Fig. 3.4-8c) is an alternative to riveting nuts. Frictional heat is generated by a multi-polygon that is pressed onto the face sheet in axial direction at a
high revolution rate/speed. The material plastifies and becomes easily formable.
One will get a defined drill-hole and the material of the face sheets will flow
into the core. The length of the formed hole wall will be three to five times the
thickness of the face material. The minimum revolution rate for flow drilling
lies at about 2400 rpm at a spindle moment of 1.5 kW. A coated thread cutter is
used at a revolution rate of 500 rpm. The lifetime of the flow former and the thread
cutter amounts to 10 000 drills and threads. But so far only feasibility studies have
been carried out.
3.4.3.6
Riveting
AFS parts may also be riveted. Especially in mixed constructions, riveting is very
well suitable. It is important to choose a relatively large diameter of the rivet
head since too small a diameter leads to a fastening pressure of the rivet that compresses the core. Due to the high surface pressure one will find plastic deformations on the face sheets.
3.4.3.7
Bonding
AFS with face sheets of aluminum may be bonded with the same technology as
conventional aluminum sheets (Fig. 3.4-9). The same parameters have to be respected as there are the creation of a defined surface, a construction adapt for
bonding and especially the choice of an adequate bonding system. The strength
of today's bonding systems partly exceeds the physical values of the aluminum
foam cores. One of the main advantages of bonding lies in the optimum transmission of the applied forces. This technology allows the AFS parts to be excellently
integrated into the surrounding structure.
3.4.3.8
123
124
Bonding of AFS
3.4.4
Cutting of AFS
Cutting of AFS structures, the unfoamed precursor material as well as the foamed
sandwich panels, can be quite a challenge, especially in the latter case. Here, conventional mechanical cutting techniques cannot be applied in a straight-forward
way because of the danger of uncontrollable deformation of the material. Therefore, two alternative methods have been evaluated.
References
9. W. Seeliger, Entwicklung und Programmierung eines Materialmodells fur elastoplastische Metallschaume, Thesis, University
of Bremen, MIT-Verlag, Bremen 2000.
10. W. Seeliger, Fertigungs- und Anwendungsstrategien fur AluminiumschaumSandwich Bauteile, AFS (Aluminum Foam
Sandwich) in Proc. IIR Werkstoff-Kongress,
22 23 Jan 2001.
11. R. Braune, A. Otto, in Metal Foams
and Porous Metal Structures, J. Banhart,
M. F. Ashby, N. A. Fleck (eds), MIT-Verlag,
Bremen, 1999, p. 119.
12. R. Braune, A. Otto, Materwiss. Werkstofftechn. 2000, 31, 562.
13. R. Braune, C. Beichelt, in Lasertechnologien
im Automobilbau, Stuttgart, 29 Feb 1 Mar
2000.
14. C. Beichelt, in Metallschaume Verfahren
und Komponenten fu
r den Leichtbau, J. Baumeister (ed), Workshop, Bremen, 7 8 Mar
2001.
125
4
Characterization of Cellular Metals
B. Kriszt
We are familiar with natural cellular structures such as wood and bone: these natural materials are structured hierarchically.
In engineering, man-made cellular materials (porous construction materials and
cellular polymers) have been used for a long time because of their remarkable thermal insulation. A new group of cellular materials has been invented more recently:
the cellular metals. When we study different man-made cellular materials we soon
recognize that they also show a hierarchical architecture. As in nature, every production process leads to a different cellular structure, so clearly cellular structures
are determined by the process by which they are made. However, when we want to
describe a cellular metal we say Alporas, Alulight, or Inco foam, and do not describe exact structural quantities.
So before we can characterize cellular metals, we need to understand the significant parameters necessary to describe the architecture and the hierarchy.
First, we can think of a cellular metal as a heterogeneous composite material of a
metal and a gas. If only the metal matrix is observed, the microstructure consisting
of grains, precipitates, dendrites, or different phases can be seen, but there is no
information on the architecture of the cellular metal. This hierarchical level is
often called the microstructure of the metal.
The density of the cellular metal depends on the volume fraction of metal and
gas. Consequently, cellular metals are characterized by their density compared to
the parent metal: the relative density. Typical metal foams have relative densities
of less than 0.3. Typical porous metals have relative densities greater than 0.3
but less than 1.The density of cellular metals, in absolute units or as relative density, is one of their main characteristics. From the density of a cellular metal part,
we have an instrument for describing the material at the macro level.
The second main characteristic is the cellular architecture. Often fluctuations in
cellular architecture lead to inhomogeneity in density, so the local density distribution can describe a cellular metal more precisely. This way of describing cellular
structures belongs to the meso level.
127
128
However, the definition of macro, meso, and micro is not yet standardized: in
Section 4.1 a scheme is discussed that helps to clarify the classification of the hierarchical structure.
Real cellular metals do not have perfect ideal cells, but the analysis of ideal cells
is nevertheless helpful for describing the architecture of cellular metals. The ideal
cells model describes the whole structure in terms of at least one space-filling
three-dimensional unit cell; in most cases we can think of polygonal cells. For example, honeycombs are networks of channels having hexagonal cross section; metallic foams, sponges, or lattice block material are representatives of real 3D polygonal cellular metals.
Cellular structures in which adjacent cells are separated by cell walls or faces are
called closed-cell foams or metals: the other group, open-cell foams or metals, are
permeable. Sometimes this classification is not clear, because in some cellular metals both closed cells and open cells can be observed.
To get a feeling for a 3-D structure, we conceptualize a space-filling unit cell of
an ideal structure (Fig. 4-1). From a geometric point of view it is helpful to think of
a cellular structure formed by a skeleton of cell edges (struts), which merge at
nodes (vertices).
One space-filling unit cell is the Kelvin tetradaidecahedron cell. This cell has a
low surface area per unit volume. Most foams made by expanding a gas in a liquid
tend to form such a cell shape. For this reason, this cell type is often chosen for
modeling cellular metals [1 6]. Knowing the geometry of the unit cell allows us
to calculate the relative density of the cellular metal. From the example of the Kelvin cell, we can elaborate the main structural features used for characterization.
Certain parameters are needed to describe the cell shape, the cell size, the topology, the geometry of cell struts, nodes, and walls, and their relation to neighboring
cells. Often it is helpful to define also parameters that describe the imperfections
or the defects of a cellular material compared with the ideal unit cell (see
Section 4.1).
A long list of parameter can be defined, but we have to ask critically if these serve
a useful purpose. If we want to understand relations between processing parameters and cell structure (Chapter 2, and Section 4.1), as well as between cell structure and the property profile (Section 4.1 and Chapter 5), we need to investigate
and quantify the cells and their imperfections as precisely as possible.
Another task that requires structural parameters is modeling the mechanical behavior of cellular metals (Section 4.1 and Chapter 6). Input data derived from characteristic cell features of real foams are needed for generating virtual cellular networks, such as the Voronoi network, for examples.
References
129
130
4.1
It is known from modeling of ideal cell structures and demonstrated by experiment, that the real cell structure of a cellular material is the key factor that determines its properties. This fact leads to the need for characterization of cellular materials. The objective of characterization is to define the relevant structural parameters and to understand their influence on the properties of the cellular material.
In this section the main structural features are described and their influence on
properties is shown.
4.1.1
As can be seen in Fig. 4.1-1, we can define a list of more than 30 structural parameters for describing the geometrical structure (architecture) and the microstructure of cellular materials.
As already discussed in the introduction of Chapter 4, cellular metals are heterogeneous materials, like composites, because the pores are embedded in the metal
cell skeleton. The shape and arrangement of the cells is called the geometrical
structure or architecture. In some cellular metals the geometrical architecture
List of parameters for describing structure of metallic foam; depending on the resolution, geometrical structure and microstructure can be defined.
Figure 4.1-1.
can be observed visually, so the length scale of cell size can range from some tenths
of millimeters to some centimeters. For acquiring images of the outside appearance of the geometrical structure either digital video cameras or ordinary scanners
can be used. Typical magnifications that can be reached with these systems are up
to tenfold.
If we zoom in on the cell material, we will see another hierarchical level or substructure, the microstructure of the cellular skeleton. We can observe the structure
of metallic cell material, micropores, and cracks at a resolution of microns by using
light microscopy or scanning electron microscopy. If transmission electron microscopy is used, the lateral resolution is increased to nanometers and we may observe
dislocation structure or small precipitates.
From this, it becomes clear that cellular materials are more than heterogeneous
materials: they are structured hierarchically. When describing cellular materials,
one has to be aware that different structural features that can influence the properties of the cellular material can be defined for each hierarchical level.
Understanding such a complex system is rather difficult. As a consequence, no
general material rules based on real structures can be developed to explain the influences of structure and architecture on properties. Furthermore, no guidelines
for the characterization of cellular metals have been established. In recent years
structural features that influence the properties have been identified but their interactions and effects on real metal foams are not known quantitatively.
131
132
Figure 4.1-2.
the same density, but one samples can have one or two big pores that represent
most of the volume fraction of the gaseous phase, and the other sample (for instance made by a replication technique or investment casting) can consist of a network of nearly uniform small pores. Figure 4.1.-3 shows differences in structures
of cellular material.
a)
Figure 4.1-3. a) Closed-cell foam, geometrical structure of Alulight foam produced by PCF: irregular pore size and
shape can be seen.
b)
The impact on properties is different; it has been shown that cell size of carbon
foams has an influence on compression strength, but not on stiffness and fracture
toughness [6].
Even if we assume that pore size has an impact on the mechanical behavior, no
experimental work shows how large pores act in real metallic foams, so they have
to be classified as a knock-down factor, or defect, for mechanical properties. It
has been shown that the size of the pores, combined with a certain shape, for example an elliptical shape and a certain orientation, can allow deformation bands to
start [7]. In Fig. 4.1-4, orientated and elongated pores in Alulight are marked. A
study of pore characteristics in Alulight foam revealed that pores in the deformation zone have significantly different orientation from pores in undamaged material [8]. Similar behavior was deduced from in-situ experiments using computed
tomography [9]. Unlike structural applications, the pores size becomes an important feature if the cellular metals are used in heat exchangers or in filter systems
(see Section 7.2).
133
134
Cell Skeleton
After characterizing the pore structure, it still remains to define quantities for describing the cell skeleton. As explained in the introduction of this chapter, the skeleton consists of cell nodes interconnected by cell edges. In a closed-cell structure,
the cell edges are connected by cell walls that separate the cells.
The volume fraction of metal in nodes, edges, and walls defines the density of
the cellular material. The features that can be described are number of nodes,
number of cell edges and walls per cell, length, thickness, and curvature of cell
edges, angle between nodes and cell edges, and shape of cell edge, number of
cell walls connected of edges and of edges connected at nodes (connectivily). A
drawback of quantitative image analysis of nontransparent samples is that it is
difficult to define the exact position and extent of the cell walls, edges, and
nodes, as well as the range of these features.
All these parameters have an influence on mechanical properties. As long as the
cellular material is nearly uniform and perfect, these are the most important parameters, but in real foams parameters are necessary to characterize agglomeration
of solid material. Klocker proposed the analysis of covariance function for detection
of clustering [10]. In a comparison of undeformed and crushed Alporas samples,
clustering of cells could be shown by a drop in covariance. Another way of characterizing clustering is the pair correlation function [16], but this method has not yet
been applied for describing cell clustering in metallic foams.
Defects in Cellular Metals and Foams
Another important aspect of characterizing the cell material is the description of
imperfections. All structural features that degrade the properties determined by
the scaling law of Gibson and Ashby, are classified as knock-down factors or defects.
First of all there are geometrical imperfections. These are deviations from ideal
polygons or irregularities in edge pattern. Models (see Chapter 6.1) revealed that
imperfections or defects such as wavy distortions of cell walls, variations of cellwall thickness, or non-uniform cell shape decrease the stiffness and strength of
cellular material [11 14]. Unfortunately, most metallic foams tend to show many
of these defects.
Defects in cell edges and walls have been identified as reducing the elastic modulus of metallic foams significantly. Two different typical defects are known: curved
cell edges and corrugated cell walls and edges. Corrugated cell walls are typically
found in metal foams such as Cymat (Alcan) or Hydro foams, which have rather
thin cell walls compared to the cell size (Fig. 4.1-5). Curved or corrugated cell
walls occur during forming or on solidification.
In modeling, cell edges or walls are often described by beam or shell elements
having an ideal shape and constant thickness. In real foams, the metal is thicker
at cell nodes and edges because of the dynamics of metal flow during foaming.
Flow of the melt also causes the rupture of cell walls. During solidification, dendritic growth at a free surface and shrinkage of the melt can induce roughening of cell
walls, edges, or nodes. Figure 4.1-6 shows the surface of a solidified cell wall. The
shrinkage of the melt during solidification and cooling leads to cell-wall wriggling
and also cracks in cell walls. The influence of initial cracks on toughness or fatigue
behavior is still under discussion.
The formation of micropores is also connected to sucking of liquid into the
nodes (Plateau effect). Owing to the increased amount of liquid in nodes, the
solidification starts at the surface of the nodes caused by the heat flow. While the
135
136
nodes solidify the melt is consumed and, as know from casting, small voids occur
in the middle of the node. This can be observed as micropores in cell nodes or
thicker cell edges.
4.1.2
Although XCT is now established for the investigation of metallic foams (see
Section 4.2), the characterization of metallic foams is usually performed using
quantitative 2D image analysis methods. Most papers concentrate on the characterization of cell structure derived from analysis of cross sections. For further reading
on the principles of stereology the reader is referred to the basic literature of Exner
[15] and Ohser [16].
In the following section the structure of cellular metals and metallic foams are
described and typical structural features are discussed. Before going into detail on
structural analysis, some information on the preparation techniques for metallic
foams is given.
Sample Preparation
The first step in preparing a metallographic specimen is to cut a sample from a
foamed part. In the early days of foam research, great uncertainty led to insistence
that metallic foam should only be cut using spark erosion machines in order to
prevent damage to cells or cell walls. Our results have shown that using a precision
saw at low cutting speed and force gives results just as good as spark erosion, but
with a reduction in time and cost.
In the next step, the specimens are often embedded under vacuum in epoxy, so
that the pores are filled and a conventional metallographic preparation with
grinding and polishing can follow. The procedure of filling epoxy in the pores
has be repeated several times until all the pores are filled. This method gives
good results if the microstructure of the cell-wall material is studied using light
microscopy. If using scanning electron microscopy on epoxy-embedded samples
the surface can become charged: a higher conductivity embedding substance,
such as Cu-enriched epoxy, will help to improve the resolution and quality of
the images.
Another method of preparing the sample for digitizing by scanning was proposed by Kriszt et al. [17]. In this method the foamed samples are precision cut,
followed by coloring the sample and smooth grinding, so that the contours of
the cell structure on top of the sample regain their metallic luster. This is a practical way of analyzing the geometrical structure of big areas. After the preparation,
the surface can be scanned using a high resolution mode and converted into a
digital image. Figure 4.1-7 depicts prepared cross sections of Alporas foam prepared by this technique.
4.1.2.1
Figure 4.1-7.
Pore Size
Figures 4.1-3a, 4.1-4, 4.1-5 and 4.1-7 depict the differences in structure between
metal foams produced by the powder compact technique, such as Alulight or Fraunhofer foam, and the melt process route, such as Alporas, Cymat, and Combal Foam.
Alulight seems to have more irregular cell shape and size than Alporas. This is a
typical feature of PCF materials. Another structural feature of metallic foams such
4.1.2.2
a)
137
138
a)
b)
as Alulight is the non-uniform distribution of solid metal in the foamed part (see
Fig. 4.1-3a and 4.1-4). Zones having an enrichment of solid material are referred as
hard spots in the literature [18].
Measurements of the pores size for both foam types have shown that they have
the same average equivalent diameter of about 3 mm. The distribution curves of
equivalent diameter versus frequency of pores of both foam types show a bimodal
shape (Fig. 4.1-8a and Fig. 4.1-9a). The large fraction of small pores is caused by
solidification shrinkage. Interestingly, it seems that the Alulight material always
shows a broader distribution curve. Because of the high number of small pores,
the cell-size distribution is dominated by the small pores. The presentation of
area fraction of a pore class over equivalent diameter helps to emphasise the contribution of big pores to geometric structure (Fig. 4.1-8b and Fig. 4.1-9b). The
presence of outliers can be made more visible if this method of presenting the results is used. Outliers having a pores size bigger than five times the average pore
size can be detected in nearly all samples produced by PCF techniques, such as
Alulight. This is not the case in metallic foams produced by the melt route.
Figure 4.1-10. Density as a function of pore size for various foam types (data for FORMGRIP
process from [20]).
Pore Shape
Assuming that cells of metal foams have ideal polyhedral shapes, such as the Kelvin cell, it should be possible to find a correlation between that number of sides of
each pore in the 2D cross section and 3D shape [15,16]. But Degischer et al.
showed that the pores in metallic foams do not have an ideal polyhedral unit
cell [18]. Hence, in 2D analysis the characterization of shape is limited to the
use of shape factors for comparison of different foam types.
4.1.2.3
139
140
From quantitative structology different methods for defining shape factors are
known. Klocker defined the ratio between the largest and the smallest principle
cell dimension as the shape anisotropy ratio [10]. A similar definition was used
by Simone (ellipse shape ratio) [22]. Another shape factor is circularity, which is
the ratio of pore area and the square of pore perimeter [23]. For all mentioned
shape factors a value of 1 corresponds to a circular shape. The circularity allows
one to interpret the shape of the cell; the circularity of a rectangle is 1.4 and
that of a regular pentagon is 1.1.
Klocker has shown that the shape anisotropy ratio of Alporas and Alulight has an
average ratio of 1.5 [10]. The ellipse shape ratio of Alulight foam is Gaussian distributed, and has an average value of 0.6 [17]. In Cymat foam an ellipsoidal shape
ratio of a factor of 1.8 was found [22]. Studies of evolution of pore shape during
foaming revealed that foams having a low expansion rate (high density) tend to
have pores with a circular shape. As the foaming process proceeds the shape of
the pores becomes rectangular or ellipsoidal [23].
Pore Orientation
The definition of pore orientation is only sensible if a reference system is defined.
Depending on the target of the investigation, reference axes are oriented to some
main directions of processing, such as the foaming direction or the feed direction
of conveyor-belt systems. Another kind of reference system can be defined by
the direction of loading. The only foam types in which the pores have a systematic
and significant orientation in relation to processing are Hydro and Cymat [22].
Generally, Alulight and Alporas foams do not have a pronounced pore orientation
[17,22].
4.1.2.4
Topological Features
The most difficult task in characterizing the foam structure is the description of
topological relations. If no model for the complex geometrical structure of real
foam is known, our understanding of topology of cellular material based on a
2D analysis is limited. First attempts have been made by calculating tessellation
for cell structures.
Klocker measured the average number of neighbor cells for Alporas and Alulight
[10]. The number of neighbor cells varies between 1 and 10. The average number
of neighbor cells in Alporas is 4.5, in Alulight material the average number is 5
cells, slightly higher than in Alporas. In deformation bands caused by compression, the analysis of neighbor cell zones gave a smaller number [10].
Another parameter that helps to describe the topological relations is the distance
to nearest neighbors. For Alulight foam Mosler et al. found a mean average distance of about 900 mm [24].
The characterization of the pattern of 2D cell arrangement can be described by
tessellations. Some information on tessellations of cellular structures can be found
elsewhere [10,24]. Klocker's results are optimistic about the description of Alporas
foam by Voronoi tessellations, whereas the results on Alulight, because of its
strong dependence of cell structure on the position in the specimen, are not so promising. Klocker suggested the Johnson Mehl tessellation as an alternative that
takes into account the curvature of cell walls. In contrast to Klocker's work, the investigations of Mosler et al. confirmed that Voronoi tessellation give good agreement with the geometric structure of Alulight [24]. The explanation for the contrary
results might already be given by Klocker, who assumed that inhomogeneity in the
structure of Alulight causes this result.
4.1.2.6
4.1.3
Knowledge of the microstructure (grains, grain boundaries, precipitates, dislocations etc.) of metallic foams is essential, because of the effect of microstructure
on mechanical properties. As has been shown, foamed samples having the same
density but made of different alloys reach different plateau stresses [25]. Different
hardnesses support the theory that the microstructure is responsible for different
plateau strengths [26].
More ductile alloys, such as wrought aluminum alloys, show a ductile collapse of
cells under compression; brittle cast alloys deform by crushing of cells. This different behavior is why the strength of a foam made of a wrought aluminum alloy in-
141
142
Figure 4.1-11.
Figure 4.1-13.
Another interesting result was found by Muller et al. when studying the microstructure of Fraunhofer foams made of the casting alloy AlSi7 [30]. They found
some Ti-rich particles after foaming (Fig. 4.1-13). It can be assumed that these
particles are residuals of the TiH2 blowing agent, which are stabilized by the
formation of a diffusion barrier. Braune et al. (Section 3.1) reported that
they observed refoaming in some foamed samples when the samples were
heated before hot forming. Both observations indicate that due to the formation of intermetallic layers surrounding the TiH2 particle, it is likely that the
dissociation decelerated.
4.1.4
Conclusions
143
144
But with the ongoing research on cellular metals the investigation of microstructure is important. Often there are typical defects, such as irregularity in cell shape,
cracks in cellular structure, or micropores caused by the intrinsic properties of the
massive metal. So in future, the emphasis in foam research will be put on the
microstructure, because not all material properties can be explained by the
architecture of the cellular metal and knowledge of microstructure is essential
for understanding processing.
References
4.2
E. Maire
145
146
a)
b)
a) Diagram of the principle of the X-ray radiography technique used in the present
chapter with a diverging beam for a medium resolution. b) Diagram of the X-ray tomography
technique.
Figure 4.2-1.
Z4
E3
(2)
where K is a constant, r and Z are the density and the atomic number of the investigated material, and E is the energy of the incident photons. Above 60 keV
the Compton effect dominates, where the attenuation becomes nearly proportional
to the density.
This attenuation law explains the contrast observed in the X-ray radiograph of a
bulky material (see Fig. 4.2-1a) because each point of a detector placed behind the
sample is situated in front of a different path: if the material is heterogeneous, the
integral value of m(x,y,z) varies also with x and y.
X-ray Tomography
The drawback of radiography is that a large amount of information is projected on
one single plane and the resulting image can be difficult to interpret if the number
of microstructural features along the thickness of the sample is important. Tomography overcomes this drawback by combining the information from a large number of
such radiographs, each being taken with a different orientation of the sample in
front of the detector (Fig. 4.2-1b). If the angular step between each radiograph is
small enough, it is possible from the complete set of radiographs to recalculate
the local value of m(x,y,z) at each point of the sample. This reconstruction is performed by appropriate software based on the filtered back-projection algorithm [1].
4.2.1.2
4.2.2
Set-ups
Different set-ups can be used. They all include a source, a rotation stage on which
the object is fixed, and an X-ray detector. The easiest way of getting digitized
images is to use directly a two-dimensional radioscopic detector composed of a
screen transforming the X-rays into visible light, which is then transferred by appropriate optic lenses to a cooled CCD camera. The common geometric constraint
of the different set-ups is that the axis of rotation of the sample must be parallel to
the plane of the detector and its image must be aligned with one of the columns of
the CCD (preferably the central column). The crucial point in applying tomography
to materials science is the achievable spatial resolution r defined as the size of the
pixel of the detector in reference of the sample. Its limit value is mainly governed
by the available photon flux at the level of the sample and by the set-up, as described in the next two sections. For medical tomography (XCT scanner) this
limit of resolution is of the order of 300 mm. Materials scientists wishing to see
and distinguish features with a size of the order of 1 10 mm had to develop appropriate tools.
Medium-Resolution Microtomography
For a limit of resolution of the order of 10 mm (medium resolution), a cone beam
system can be used with a classical microfocus X-ray tube as the source. Such a
device has been assembled in the CNDRI laboratory at INSA Lyon. With this diverging geometry the magnification can be varied easily by changing the position of
the sample in the space between the source and the detector. The limit of resolution is due to the size of the microfocus, which introduces blur in the projected
image. This size has a minimum value because if the source size gets too small,
the flux at the level of the sample becomes so low that the acquisition time required to record a single radiograph is too long for a realistic analysis. In such a
laboratory set-up, a polychromatic source is used in order to keep the acquisition
time acceptable. This may introduce artifacts due to beam-hardening and does
not allow quantitative reconstruction of the absorption coefficient m.
This kind of standard set-up exists now in many different academic and industrial laboratories. Some commercial and nearly portable set-ups with a good resolution (down to 6 mm) are also available.
4.2.2.1
High-Resolution Microtomography
We emphasized in the previous section that the set-ups using X-ray tubes are
limited because of the low flux delivered by this kind of source. The best quality
images in terms of signal-to-noise ratio and spatial resolution, allowing high-resolution microtomography, are today obtained from instruments located in synchrotron radiation facilities. Such a source can be found at ESRF in Grenoble.
The X-ray beam produced at this kind of third generation synchrotron facility
4.2.2.2
147
148
is very interesting for microtomography because of the following original features [2].
x
x
The ID19 beam line team of ESRF has combined these specific properties with the
availability of a suitable CCD-based detector exhibiting at the same time a large
dynamic range (13 bits), a low noise, and a short transfer time (down to 60 ms
for 1024 q 1024 pixels). This unique combination allows a resolution of the
order of 1 mm to be obtained routinely today.
Reconstruction Method
Reconstruction of the images presented here is achieved using a C-language program written by the CEA-LETI (Commissariat a l'Energie Atomique, France) and
based on the Feldkamp algorithm [8]. A ramp filter is selected, thus preserving
the spatial resolution. A DEC 500 MHz workstation is used to reconstruct the entire volume, slice by slice. Reconstruction of a typical volume (350 q 340 q 128 voxels) takes approximately 1 hour. The final result is a 3D map of the m of elementary
volumes of size r q r q r inside the sample. This map can give a certain image of
the microstructure if m is transformed into a gray level. Voxels located in the
aluminum will exhibit high values of m (high gray levels) whereas voxels located
in the pores will exhibit low values of m (low gray levels).
4.2.2.4
4.2.3
Experimental Results
Initial Cell Structure
Figure 4.2-2 compares reconstructed slices (in gray level) of two different Norsk
Hydro foams with different density and cell sizes. These slices are 2D images
but extracted from a 3D bloc of numerical data (i. e., the reconstructed volume
4.2.3.1
a)
b)
of the sample). The internal structure and cell-wall arrangement is clearly imaged
and the difference between the two samples is easy to visualize. The relative densities of the two materials shown in these figures are 6 and 11 %. A 3D rendering
of the 6 % foam is shown in Fig. 4.2-3. This rendering is obtained by removing the
pixels located in the pores, showing then only the material located in the walls of
the foam. This virtual image really looks like an optical picture of foam, as can be
seen in different sections of this book. From such images, 3D morphological characteristics can be measured such as the wall thickness, the wall dimensions, the
cell size and so forth. These characteristics have a strong influence on the mechanical properties.
Note also that these images can be used with a lower resolution to assess the
density fluctuations inside the material (m is directly related to the density). It is
perfectly well established that the value of the global relative density of a sample
of metallic foam has a direct effect on its mechanical properties but the density
fluctuations have also a strong influence and especially on the scatter of these properties. This aspect is developed in Chapter 6 of this book.
149
150
a)
b)
c)
d)
Figure 4.2-4. a) Two dimensional reconstructed slice showing the initial microstructure of the compressed sample.
b) Sample slice as in (a) for e 0.065.
c) Sample slice as in (a) for e 0.2.
d) Sample slice as in (a) but e 0.6.
The compression set-up used for this study was an Adamel Lomargy standard
mechanical frame. The sample was placed between two plates and the force/displacement curve was recorded. The resulting s(e) conventional curve was found
to be classical for this material with a yield stress of 0.7 MPa and a slight increase
in the stress during the plastic collapse. We made three steps during the compression loading to allow the characterization of the internal structure by tomography.
The three steps were made at increasing values of the remanant plastic true strain:
0.065, 0.2, and 0.6, measured directly from the change in dimensions of the parallelepiped after unloading.
Figure 4.2-4 shows a set of 2D reconstructed slices of the same zone of the sample numerically and nondestructively extracted from the volume at the four compression stages from 0 to 0.6 of remanant true strain. These slices are parallel to
the compression axis, which is vertical on the figure. One can clearly see the deformation mechanism of the studied foam in compression by comparing figures (a)
(initial state) and (b) (remnant strain 0.065). The large value of the plastic strain
at this stage is clearly not due to a homogeneous plastic straining of the whole
sample, but to the local buckling of several walls. Between the second and the
third step, we observe that the number of new buckled walls is small. The deformation process leads first to the complete closing of the collapsed cells instead
of the appearance of new ones. From the observation of the entire population of
buckled walls, we have observed that they are situated in a band perpendicular
to the loading axis (see Fig. 4.2-4c). These 3D nondestructive observations confirm
those made using other techniques [9] or with the same technique by other authors
[3,5,6].
4.2.4
The internal structure of a metal foam is extremely complex: the cell size and the
length of the walls are parameters that are widely spread. The nonperiodic character of the structure leads also to a very difficult calculation of the actual force applied to each wall compared to the case of honeycombs. The complexity of this entanglement suggests analyzing the distribution of the stress and deformation in
each wall using an appropriate tool: finite element analysis. In the present study
we have a 3D picture of the actual microstructure of the foam. In what follows,
we present an easy way to use this picture to generate meshed models of the actual
microstructure. These models are readable by the Abaqus commercial code, which
we have used to perform calculations, firstly in 2D and then in 3D.
151
152
a cubic element in a finite element mesh. The cubic elements located in the voids
are then removed from the model. Given the experimental conditions we used, we
considered that the boundary between the steel plates of the compression rig and
the aluminum foam was perfect. The displacement of the lower nodes were fixed
in the three directions. We generated a steel plate perfectly bonded to the upper
nodes and applied a displacement along the compression direction to the upper
nodes of this steel plate.
Results
All the calculations were performed assuming elastic behavior for the walls.
4.2.4.2
Two-dimensional (2D)
The procedure described above can be very simply applied to any slice by generating only one layer of four-node 2D solid elements. We applied this technique for
several slices. The elements used were plane-strain four-nodes elements (PE4 Abaqus elements).
Figure 4.2-5 shows a contour plot of the s22 (i. e., the normal component in the
direction of loading) component of the stress tensor in the walls calculated according to our procedure for a slice with PE elements. The higher compressive stresses
are shown in black on the figure. What we want to reach with this calculation is not
the absolute value of the stress but the determination of the sign of this stress so
the color bar has been intentionally simplified to distinguish tension and compres-
Figure 4.2-5. Stress map (arbitrary units) calculated using the Abaqus FE code by converting the
actual microstructure of a 2D slice into a mesh using the local procedure described in this
chapter. The stress component shown here is the normal component.
Enlargement of a region of the stress map shown in Fig. 4.2-5, which shows the
bending character of the stress field when calculated in 2D.
Figure 4.2-6.
sion. Note that the values of the stresses in the walls seem to form paths of high
compression loading between the top and the bottom of the section. These paths
define pillars where the compression load is transferred while some other regions are much less loaded. This permits us to locate highly loaded regions in
the material.
If we zoom in a single wall situated in one of these pillars and oriented along the
compression axis, we observe a very special stress state composed of zones in compression (black) and in tension (light gray), which indicate a high bending character of the stress (see Fig. 4.2-6) with 2D assumptions. Plane-strain elements imply
that the cells that are defined in Fig. 4.2-6 are section of tubes having an infinite
dimension along the third axis perpendicular to the figure. This tube shape promotes a high bending stress state within the plane of the slice only.
Three-dimensional (3D)
A complete 3D calculation of the entire foam with our technique would require a
1 000 000 element mesh. This amount of data is not tractable with our computer
today. A region of 1003 voxels was selected from the initial raw data. In order to
reduce the number of elements in the final mesh, the voxels of the region were
grouped eight by eight (two by two in the three directions), which led to an increase
in the actual resolution in the final mesh leading to a size of 300 mm q 300 mm q
300 mm for each element. This 503 voxels 3D block was transformed into a mesh
according to the previously described procedure and the mesh was slightly compressed (0.02 % of elastic strain). The calculation took around 140 h on a standard
workstation. This allowed us to analyze the stresses and strains in the walls as illus-
153
154
trated in Fig. 4.2-7 and Fig. 4.2-8, which show a contour plot of the normal stress
and of the Tresca shear stress in a layer. As for 2D simulation, the color map has
been simplified to emphasize the sign of the stress rather than its absolute value.
Note that Fig. 4.2-7 shows much less region loaded in tension (black) than the
equivalent 2D calculation, shown for instance in Fig. 4.2-6. This indicates that
when calculated in 3D, the stress state in the walls is completely different than
Figure 4.2-7. Stress map (arbitrary units) cal- applied strain was 0.02 %. The amount of tenculated using 3D elements on a representative sile stress is much lower than for the 2D calregion of the actual compressed block. The
culation (compare with Fig. 4.2-6).
normal stress component is shown here. The
Figure 4.2-8. A map of the value of the Tresca's shear stress calculated in the same 3D conditions as for Fig. 4.2-7.
Conclusions
Tomography is a new powerful technique, suitable for the 3D nondestructive analysis of the mesostructure of metallic foams and its evolution under compression.
We have presented illustrations which show that it is possible to use this technique
to have a better determination of the microstructural features responsible for the
mechanical behavior of such materials. When used at different compression stages
on the same sample it also enables us to understand perfectly the deformation
mechanisms of metallic foams. In the case of the Norsk Hydro foams, we have
shown that deformation is due to local buckling of the edges and walls and that
these events were all located in a band perpendicular to the compression axis.
These images also give directly produced finite-element meshes of the actual samples. The size of these meshes can be very big and not tractable with a regular
computer if the entire compression samples are meshed in 3D. Preliminary calculations were then made on entire sections in 2D, which allow us to visualize the
highly loaded regions in some particular sections and the bending character of
the stress state in the walls, but the local carrying function of the cell edges cannot
be taken into account appropriately. A small 3D block was finally calculated and it
was thus shown that the stress state in the walls has a much less pronounced bending character in the actual conditions.
References
1. J.-Y. Buffiere, E. Maire, P. Cloetens, G. Lormand, R. Fougeres. Acta Mater. 1999, 47,
1613.
2. V. Kaftandjian, G. Peix, D. Babot, F. Peyrin,
J. X-ray Sci. Technol. 1996, 6, 94.
3. H. Bart-Smith, A. F. Bastawros, D. R.
Mumm, A. G. Evans, D. J. Sypeck, H. H. g.
Wadley, Acta Mater. 1998, 46, 3582.
4. E. Maire, F. Wattebled, J. Y. Buffiere, G. Peix,
in Proc. Euromat `99 Conf., Vol. 5, Munich,
T. W. Clyne, F. Simancik (eds), Wiley VCH,
Weinheim 1999, p. 68.
5. H. P. Degischer, A. Kottar, B. Foroughi,
Ch. 12 in X-ray Tomograhy in Material Science,
J. Baruchel, J.-Y. Buffiere, E. Maire, P. Merle,
G. Peix (eds), Hermes, Paris 2000.
155
156
4.3
The quality of a material is defined by its properties with respect to the expected
performance. A general demand for engineering materials (not necessarily for objects of arts) is reproducibility in the sense that the product exhibits the same properties within specified limits delivered in different samples or batches in order to
be predictable in its performance and an engineer can design a reliable component
suitable for series production. Material science relates properties to microstructural
features of bulk materials, for which quality criteria can be deduced defining limits
for structural variations (for instance grain size) and for inhomogeneities (for instance frequency and size of inclusions). Usually the processing steps producing
the microstructure are specified to guarantee the correlated properties.
4.3.1
Introduction
Cellular metals are a new class of materials on the brink to industrial application
and therefore the producers have to develop specifications for their product with
respect to the application aimed for. In Chapter 7 various applications are proposed, each requiring different property profiles. Applications in architecture obviously will require different quality criteria than heat exchangers or crash absorbers. A design guide [1] based on the available properties of different cellular metals and material laws has been offered to the engineers for product development.
The great variety of processes and products presents a considerable challenge for
the selection of the appropriate cellular material for a potential application.
4.3.2
Regular cellular structures like those considered for modeling [2] can hardly be produced. The most regular 3D structures are achieved in lattice block material (LBM)
[3], which can be simulated by a 3D periodic array of volume elements that represent the main structural features and exhibit the corresponding properties. The behavior of any macroscopic part made of LBM can be calculated (see Chapter 6) by
means of known properties of the representative volume (RV), which contains just
as many cells that the volume related properties do not change by adding more
cells, but would change if one cell were taken off. RV is the smallest volume exhibiting the properties of the corresponding cellular structure of a given material
with a certain cell architecture yielding an uniform average density. Consequently,
no dimension of a component should be smaller than the RV, if the performance of
the specified cellular structure were required. Any disturbance in the regularity of
157
158
Figure 4.3-1.
their specific loading conditions related to the geometry and to the surface. In a
first approach, the cellular metal or part is defined by its processing route, although
the interaction of the different processing parameters cannot be fully correlated to
the produced cellular structure yet. Each type of product may be characterized by
investigating the statistics of achieved properties and the correlated structural features accessible to quality control [7]. The range of scatter of properties acceptable
for the envisaged service performance has to be defined, which can be related to
the scatter of structural features. Beyond specification limits, defects can be defined, which have to be weighted in their influence on the service properties. According to the actual development stage, it can be expected that some defects could
be eliminated or considerably reduced by improved processing techniques.
A hierarchical approach is proposed for defining quality criteria of cellular metals by addressing different description levels according to the characterization levels described in Section 4.1: macro-, meso-, and microscopic. Corresponding features are to be defined that can be quantified and validated with respect to material
properties.
The average density of a part or component is an example of a macroscopic level.
The question is: How reproducible is the density of a given part in series production? How much variation is tolerable with regard to the scatter of related properties? Even for a given average density of a part the properties might scatter owing
to the local scatter of mass densities within the part, which may vary on a mesoscopic scale. Going into further structural details on the microscopic level, the variations in shape factors, in dimensional and microstructural features of the cell
walls and nodes have to be quantified statistically and assessed with respect to tolerable variations or irregularities beyond.
4.3.3
Macroscopic Parameters
All those parameters are referred to as macroscopic, which can be assigned to the
whole cellular part without looking into the structure in detail: the general appearance of a cellular architecture, the composition of the ingredients, and the processing method.
159
160
In Section 7.3 the problem of deforming the cell walls during mechanical cutting
operations is addressed. The degree of such deformations could be of importance
for following joining operations. The enlargement of outer surface in comparison
with a carefully prepared cross section could be employed to quantify the smearing
of cross sections of elements of the cellular structure by machining.
Foamed and cast cellular metals are usually surrounded by a metal skin. The
oxide content will be much higher than on interior cell surfaces, because that
skin was exposed directly to the atmosphere in the liquid state and during solidification and cooling. The roughness depends greatly on the used mold and its coating and may be qualified similar to sand casting quality. Attempts to measure the
thickness of the skin by eddy current tests did not succeed so far. A skin region of
about 1.5 mm thickness can be deduced from Fig. 4.3-2 of an Alulight sample,
where the local density is almost twice of that in the center of the plate. Still the
skin cannot be regarded as solid metal, but consists of porous material.
There are as well pores open to the environment in such skins, which can be
detected by dye penetrant tests [6]. Although these pores may be in the range of
tenths of a millimeter in diameter, they form a passage to the cells below, which
might be of a few millimeters in diameter. Such open pores might adsorb humidity
and thus enhance corrosion attack. These pores prohibit chemical treatments like
anodizing, because the electrolyte penetrates and can hardly be removed. On the
other hand, if those pores are sealed by a coating, which has to be baked, it
might form blisters when the temperature rises and the internal gas pressure increases. Ambient temperature coating processes with a strong sealing effect can be
applied successfully [9]. The number of surface pores per surface area and their
distribution of diameters are quantifiable surface quality features. A pore density
of 1-3 pores/cm2 has been measured on Alulight samples [6].
The dimensional accuracy of net shape metal foam parts depends on the thermal
stability of the mold and the uniform shrinkage of the foam during solidification
and cooling. The cell-size distribution, especially along the surface of the part, affects the surface roughness. Big pores (in the range of one centimeter) may cause
dents due to the drastic reduction of pressure of the gas in the closed pores during
cooling. The achievable tolerances for net shape metal foams may be in the range
of percent, but no detailed study has been published so far.
Apparent Density
The apparent density of a cellular metal part still is the most important selection
criterion. It can be measured by the Archimedes principle provided the surface
can be sealed to avoid penetration of the test liquid. Foamed net shape parts
with a skin can be measured in high viscosity liquids preventing penetration
into the surface pores. Generally, a measurement uncertainty of about e2 % in
the apparent density of a part has to be reckoned with. The result includes the surface skin, the density and thickness of which are usually unknown. This may cause
a considerable uncertainty in the apparent density of the interior structure. The apparent density of the whole sample referred to in Fig. 4.3-2 for instance, with a
high density skin layer occupying about 15 % of its volume, is overestimated by
about 15 % (0.57 g/cm3 for the whole sample, without skin about 0.5 g/cm3).
The reproducibility of the apparent density depends on the processing technique: cellular metals produced on the basis of polymer foam space holders are
highly reproducible, but powder compact foaming achieved so far only about
e15 % reproducibility in the apparent density of similar samples from small series
prototype production. An overall non-uniform mass distribution can be identified
by determining if there is a difference in the position of the center of gravity of the
shape and that of the mass.
4.3.3.3
Properties
Macroscopic properties of test samples are described in Chapter 5, where they are
usually related to the apparent density of the specimens. The observed scatter in
mechanical properties, like the plateau stress for instance, depends very much
on the uniformity of the cellular structure. In general, the properties are closely
related to the quality of the cellular architecture. The scatter in the test results indicate the variations in structure of the individual test specimens. Most of the tests
are executed without knowing the microgeometry of the test sample exactly, therefore there is little evidence on the correlation with structural features, especially
when the tests are destructive. A systematic correlation of properties to the
three-dimensional cellular architecture is highly desirable in order to give rise to
process developments aiming for improvements in the reliability of the service
related performance of the foams.
4.3.3.4
4.3.4
Microscopic Features
In terms of microscopic features the individual cells are to be considered: the microstructure of the metal forming the cell walls and their topological parameters
described in Section 4.1.
161
162
Geometrical Features
The shape of cells and cell walls has been investigated so far by metallographic
methods either by light microscopy or by scanning electron microscopy. Average
cell-size distributions from cross sections are only representative when the pores
are spherical. The presentation of pore-volume distributions is more relevant
than equivalent linear pore diameter statistics. Average pore size figures are of little
significance because the width of the distribution determines the properties according to the weakest link principle. The distributions may be Gaussian only
for hollow sphere structures and those produced by replication techniques,
where the narrow width of the distribution provides a quality measure. Generally,
the shape of the pore volume distribution curves could provide a quality criterion
for the cellular structure. The portion of pores bigger than the average is especially
important for the properties of foamed metals. In a first approach the statistics of
over-sized pores could be used as quality criterion: for instance, the volume fraction occupied by the biggest 10 % or the pores bigger than 20 % of the sample's
thickness. Not only the size of the big pores is relevant, but as well their positions
and the microgeometry of surrounding pores. A centrally placed big pore would
not affect the stiffness of a component and the strength could be compensated
by surrounding small pores (see Chapter 6.1.3). A surface near big pore could
be harmful at the position of load transfer, where the local pore size distributions
would be of interest.
Some attempts have been made to determine the three-dimensional geometry of
pores by computed X-ray microtomography [5,12]. It gives the most exact tool for
4.3.4.2
Alulight AlMgSiCu of r 0.39 g/cm3, c) Alulight AlSi10Mg r 0.37 g/cm3, d) same material as (c), but specially foamed small cell architecture r 0.56 g/cm3.
163
164
Microdefects
Of course, pores created in cellular metals due to processing imperfections like
shrinkage pores, imperfect sintering, spraying pores etc., may not to be considered
as essential defects. But in general the consistence of the percolating solid should
be of similar metallic properties throughout the cellular structure. Cellular structures processed in the melt or at elevated temperatures suffer thermal stresses during solidification and cooling. Hot tearing and cracking is very probable because of
the multi-axial stress situations created in walls of little strength (see Section 4.1).
Almost 100 % connectivity of the pores have been determined within powder compact foamed aluminum [5], although almost 40 % of the solid is located within the
cell walls. Cell-wall cracks have been observed frequently in powder compact foams
[19].
The material properties of the metal in the cellular structure have to be considered to be of poorer quality than those of the same metal in bulk. For those produced from the melt it is expected that the alloy exhibits the properties of a low
quality casting. If the frequency of microdefects depends on the position within
a sample, their effect should be considered as deviation from uniformity.
4.3.4.3
4.3.5
Mesoscopic Features
The mesoscopic description of real cellular metals assumes that a sample or part
can be described as an arrangement of volume elements, each representing a characteristic microgeometry. The local inhomogeneities in architecture and in mass
distributions can be classified by different mean cellular structures approximating
the properties of each sub-domain of the material. A mesoscopic model for uniaxial
compression has been proposed [20,21], which divides the samples into layers of
cellular materials with different apparent densities. Generalizing this concept,
any sample or part can be divided into a 3D arrangement of sub-domains of different cellular phases of known features and properties. Thus the cellular sample is
considered as a multiphase material (see Fig. 4.3-1), where a cellular phase is
defined by its composition, microgeometry, and mass distribution occupying the
volume of a sub-domain.
The mesoscopic behavior of each sub-domain corresponds to that of the associated RV, so that each sub-domain behaves like a cellular material to which macroscopic material laws of cellular metals can be assigned. In the extreme case, where
the sub-domains were smaller than RVs (i. e., most cells were essentially different
from their neighbors), such a mesoscopic approximation would not make sense
and complex finite element modeling of the real microstructure would be necessary as shown in Section 4.2. In the best case of samples of uniform microgeometry like Duocel, Incofoam, or those cast by replication, the mesoscopic consideration is superfluous and the macroscopic treatment would result satisfactorily.
In general, the range of variations in size, volume fractions and properties of
the mesoscopic sub-domains can be quantified providing a quality measure on uniformity of the considered part.
165
166
Density Distribution
If the microgeometry of the cells were uniform, except the local variations in cell
size, the density distribution would become non-uniform. Powder compact foamed
samples usually contain bigger cells in the center of samples of a few centimeters
in thickness yielding a graded density distribution (similar to that shown in
Fig. 4.3-2). Furthermore it has to be admitted that inhomogeneities in foamed
metals (over-sized pores and local mass concentrations) cannot be fully controlled
so far. The point is to take them into account by simulation. The non-uniformity in
solid distribution of foamed metals can be characterized by inhomogeneous density distribution [6]. Regions of similar mass densities can be approximated by a
mean density value [25]; thus the sample can be divided into sub-domains of different mass densities. A procedure to produce such sub-domains will be described
in Chapter 4.3.7.
4.3.5.2
4.3.6
The features characterizing parts or samples made of cellular metals are correlated
to the structural hierarchy described above in a matrix presented in Table 4.3-1,
from which quantifiable magnitudes for quality considerations are deduced.
Most of the microscopical parameters are described in Section 4.1 and require considerable effort to be determined statistically, especially in 3D. The macroscopic
magnitudes can be obtained from the manufacturer and can be complemented
by nondestructive tests except the mechanical properties. The mechanical properties of uniform cellular metal parts can be calculated by simulating the geometrical
condition with input data from test samples of equal materials. The properties of
non-uniform cellular parts (either intentionally graded or inhomogeneous due to
the processing conditions) can be estimated from their mesoscopic structure
when the scaling laws with respect to the relevant microstructural features are
known.
Mesoscopic parameters can be defined on the condition that ensembles of cells
exhibit similar microstructural features, which form more or less uniform sub-domains of a sample of cellular metal. An attempt is made to characterize a cellular
material without determining microscopic features quantitatively. Irregularities in
a certain depth from the surface can be detected by ultrasonic spectroscopy [26] as
Microscopic
Magnitudes
Dimensions of
constituents of
cells, irregularities
Composition,
structure, thickness of outer and
inner surface
Grain structure,
phase distributions
Fraction of mass
in cell constituents
Metal properties
on microscopic
scale different
from bulk
Cell
architecture
Dimensional
accuracy
Surfaces
Metal matrix
Mass
distribution
Properties
Lateral dimensions,
radii at corners
Fabrication method,
properties, roughness measure, surface porosity index
Chemical analysis of
ingredients, thermomechanical manufacturing parameters
Absolute or relative
mass density, difference of centers of
gravity of geometry
and of mass
Experimental data
and simulation results with scatter
and validity limits
Surface formation
and topology
Alloy composition,
additives, metal condition
Average, apparent
mass density; center
of gravity
Service properties of
component
Properties of an
ensemble of similar cells
Mass distribution
coefficient between
faces, Plateau borders
and nodes
Physical and mechanical properties of solid
in function of position
in cell and sample
Quantifiability
Processing method,
cell type, permeability
Macroscopic
Magnitudes
Quantifiability
Mesoscopic
Magnitudes
Quantifiability
Matrix of features of considered quality aspects and correlated quantities within the structural hierarchy of cellular metals.
Feature
Table 4.3-1.
168
long as the cell architecture is essentially uniform. In the case of regular Duocel
samples relative density variations could be observed by eddy current measurements [27]. The following paragraphs describe the possibilities to determine mesoscopic parameters of non-uniform cellular metals and discuss their applicability for
quality criteria that can predict, if a sample's behavior complies with the specifications of service requirements.
4.3.7
For simplicity it is assumed that the microstructure of the metal and the type of
cell architecture are essentially defined by the processing method. The remaining
variables are the local density distribution and the properties of the solid material,
which have to be calibrated. A non-uniform object can be divided into sub-domains
to which the appropriate material laws can be applied to simulate the whole
part's performance. For this purpose, it is convenient to transform the cellular
structure into a continuum, which behaves like a cellular architecture (crushable
material). The experimental basis is provided by computed X-ray tomography
[13] of the 3D metal distribution of a sample, which will be locally averaged to
yield density values in each point of the sample smoothing away the originally
cellular structure.
where
g is a weight function that has to fulfill the normalization condition
R
g(u)d3u = 1. Equation (1) is also written in the form ra g*r, which denotes
the convolution of the function r with g. Cellular structures can be investigated
by X-ray radiography or by X-ray computed tomography as described in
Section 4.2. The density distribution is represented by a voxelized data set mXCT(rijk)
that can be used to calculate a density mapping as defined in Eq. (1). Due to the
limited resolution of the XCT equipment mXCT(rijk) is related to the linear mass
attenuation m(r) of the object by
Z
mXCT (rijk ) PSF(rijk s r 0 )m(r 0 )d3 r 0 PSF m
(2)
where the point spread function PSF describes the blurring of the object function
m(r) [28].
For example, the micro XCT slices in Fig. 4.3-5a and Fig. 4.3-5d are a very good
representation of the real foam (i. e., the PSF is a sharp peak), whereas the corresponding slices obtained by medical XCT (Fig. 4.3-5b and Fig. 4.3-5e) are of significantly lower resolution (the PSF is relatively broad). Since the metallic foams investigated herein consist of a single alloy, m(r) is proportional to r(r) and analogously to Eq.(2), rXCT PSF*r can be defined. Averaging these rXCT values with
the weight function w yields raaccording Eq.(1)
Z
X
X
w(rijk s rlmn )rXCT (rlmn
w(rijk s rlmn ) PSF(rlmn s r 0 )r(r 0 )d3 r 0
l;m;n
l;m;n
Z X
Z
w(rijk s rlmn )PSF(rlmn s r 0 )r(r 0 )d3 r 0
(3)
l;m;n
It should be noted that rXCT is already a function of the form of Eq. (1), i. e., rXCT
can be understood as a density mapping performed by the tomograph itself and a
169
170
Figure 4.3-6.
second mathematical averaging process reveals the final mapping ra (see Fig. 4.35c and Fig. 4.3-5f). The density mappings in this section are calculated by arithmetic mean values of the voxels rXCT(rijk) over a cuboid (see Fig. 4.3-6)
ra (rijk )
1X
rXCT (rlmn )
N l;m;n
(4)
where l, m, n run over the voxel indices of the selected cuboidal surrounding of the
location rijk and N is the number of voxels, i. e., the weight function in Eq. (3) is w
1/N within the cuboid and w 0 otherwise.
Figure 4.3-7 shows a one-dimensional model r(x) of a cellular metal with walls of
the solid density rS and embedded empty pores. Convolution of r(x) with the PSF
results in the depicted rXCT(xi) values and further averaging over intervals [xi 34,
xi 34] with N 69 yields the mean local density function ra(xi).
As long as the averaging volume of the density mapping is bigger than the resolution of the tomography, the resulting density map is almost independent of the
XCT resolution [29]. Using averaging cuboids approximately as big as the cells
suppresses microstructural details, but allows to determine local differences in
171
172
a)
b)
Figure 4.3-9. Illustration of the 3D local density distribution obtained from 4 mm q 4 mm
q 4 mm averaging volumes for a series
of six layers at distances of 3.7 mm of the same
samples as in Fig. 4.3-5: a) Alporas of
mass densities for the considered averaging volume size vary in these examples between 40 % and 160 % of the apparent density of the Alporas sample, but between
0 and more than twice of rm of the Alulight sample. The gray scale presentation in
Fig. 4.3-9 divides the samples into density classes of 0.1 rm. It is clear from the
density distribution of the Alulight sample, that no uniform material behavior
can be expected.
Representation of Non-Uniformity of Densities
Histograms of density distributions can be calculated for specified averaging
volumes, which are at least as big as the majority of the cells. Alporas shows the
narrowest scatter of local density in Fig. 4.3-10. Both types of Alulight have their
maximum frequency below the average density caused by the extended tail in
the higher density range. Isolated high density regions may not affect the mechanical properties significantly according to the weakest link principle, but would
represent a surplus of material. The shape of density histograms seems to be
characteristic for the processing method of a cellular metal [15].
Density maps can be used to quantify local density variations. The line of average
density rm and arbitrary limits of density variations of eDr are indicated in the
linear model in Fig. 4.3-7. An allowable density variation can be defined and the
volume fraction of material with mass densities below or above the lower and
upper limit, can be calculated. The 3D arrangement of volumes of extreme density
values can be presented by iso-surfaces of densities smaller or bigger than a given
limit as shown in Fig. 4.3-11. The local position, orientation and connectivity of
such hard and soft sections may be important as well for the performance of a
4.3.7.2
part [30]. If there were cells bigger than the chosen averaging volume, so that
points result in the sample where no contrast is found within the surrounding cuboid, the density mapping would contain co-ordinates of zero density. The approximate size, volume fraction, and position of pores bigger than the averaging volume can be determined by means of the calculated 3D density map. An example
for an Alulight specimen with over-sized pores is shown elsewhere [6].
X-ray radiography can be applied to flat samples. The intensity of the transmitted
columns contains the density integrated over the thickness of the sample yielding
the contrast of pixels in the projected plane. This can be transformed into 2D density maps by applying the analogous procedure of averaging over m q n pixels in
the image plane. Examples given elsewhere [6] illustrate an improved detectability
of inhomogeneities (like ensembles of big pores, drainage effects, or dense skins)
173
174
by planar density mapping. Of course, radiography does not provide any information on density variations in the direction of the thickness. A compromise between
3D tomography and radiography is provided by X-ray laminography [31], where the
density distribution of parallel layers of a sample can be determined with moderate
resolution.
4.3.8
Quality goals
Tolerances of outer
dimensions achieved
Reproducible surface
features
Defined chemical
composition
Reproducible density
Properties unambiguously
related to structural quality
Cell architecture
Dimensional accuracy
Surfaces
Metal matrix
Mass distribution
Properties
Properties, roughness,
open porosity
Specified variations in
lengths and radii
Reproducible processing
parameters
Quality criteria
Dimensions beyond
tolerances
Unsatisfactorily controlled
parameters
Defect evidence
Representative tests,
Correlation to structure,
Service performance
Density measurements,
Determination of centers of
gravity of shape and mass
Roughness measurement,
Dye penetrant tests,
Chemical analysis, hardness
Contour measurements
Quality testing
Matrix of features of parts made of cellular metals and correlated macroscopic quality criteria proposing their verification.
Features
Table 4.3-2.
Quality goals
Properties insensitive to
microstructural variations
within averaging limits
Properties of sub-domains
represented by the local
average mass density of the
corresponding volume
Cell architecture
Dimensional accuracy
Surfaces
Metal matrix
Mass distribution
Properties
Establishing confidentiality
limits for different types of
materials
Determination of location
of soft and hard regions
Calibration by experimental
tests
Definition of critical Dr
Shape, connectivity and size of
extreme densities (iso-surfaces)
Check of gradients in main
directions
Quality testing
Histogram of mass density distribution from XCT; Difference
of rm and histogram's maximum; Volume fraction of
r I rm Dr and of r i rm Dr
Defect evidence
Histogram shape, Volume
fractions below and above
certain limits
Quality criteria
Matrix of features of cellular metals and correlated mesoscopic quality criteria proposing their verification.
Feature
Table 4.3-3.
176
togram, the volume fractions of the sample with densities outside a certain specification and the local position of those extremely soft or hard regions. Even the
existence of over-sized pores can be indicated.
The following uniformity criteria are proposed for the example of Alulight
foams:
x
the local density variations should be within e30 % around the mean apparent
density;
regions of smaller mass densities than the specified range (including zero for
over-sized pores) should in any direction not be bigger than 1/3 of the thickness
of the sample;
the sample's regions with mass densities beyond the specified limits should not
be connected in any preferential orientation causing anisotropy, and should occupy less than 5 vol.-%.
Acknowledgements
References
Schwartz, D. S. Smith, A. G. Evans, H. N. G.
1. M. F. Ashby, A. G. Evans, N. A. Fleck, L. J.
Wadley (eds), MRS Symp. Proc. Vol. 521,
Gibson, J. W. Hutchinson, H. N. G. Wadley,
MRS, Warrendale, PA 1998, p. 109 117.
Metal Foams: A Design Guide, Butterworth4. J. L. Greensted, in Porous and Cellular
Heinemann, Oxford 2000.
Materials for Structural Application, D. S.
2. L. J. Gibson, M. F. Ashby, Cellular Solids:
Structure and Properties, 2nd edn, Cambridge Schwartz, D. S. Smith, A. G. Evans, H. N. G.
Wadley (eds), MRS Symp. Proc. Vol. 521,
University Press, UK 1997.
MRS, Warrendale, PA 1998, p. 3 13.
3. M. L. Renauld, A. P. Giamei, M. S.
Thompson, J. Priluck; in Porous and Cellular 5. A. Elmoutaouokkail, L. Salvo, E. Maire,
G. Peix, in Proc. Cellular Metals and Metal
Materials for Structural Application, D. S.
177
178
5
Material Properties
R. Pippan
crystal structure,
thermal expansion coefficient, and
melting temperature (solidus liquidus range).
Other physical properties, such as the heat capacity, are typically linear functions of
the density. More exactly it is the sum of the heat capacity of each phase multiplied
by its weight fraction. Even in low density metallic foams the weight fraction of gas
is small, so that the specific heat of the cellular structure is essentially equal to that
of the parent metal.
Finally, there are many properties that depend on the density (or porosity) not
only linearly, but also on the geometrical structure or the microarchitecture of
the cellular structure: the stiffness, the mechanical strength, the thermal and
electrical conductivity as well as acoustic properties, for example. It is the aim of
this chapter to give a short overview of the most important properties for the
application of cellular metals. A much more general and detailed treatment can
be found in the book on cellular solids by Gibson and Ashby [1]. The increasing
interest in the properties of metal foams is clearly visible from the increasing number of recent publications in leading materials journals [2 13], proceedings
[14 18], and the data supplied by different research groups.
179
180
5 Material Properties
Table 5-1.
Cymat
Alulight
Alporas
ERG
Inco
Material
Al SiC
Al
Al
Al
Ni
0.02 0.2
0.1 0.35
0.08 0.1
0.05 0.1
0.03 0.04
Structure
Closed cell
Closed cell
Closed cell
Open cell
Open cell
Young's modulus,
E [GPa]
0.02 2.0
1.7 12
0.4 1.0
0.06 0.3
0.4 1.0
Poisson's ratio, n
0.31 0.34
0.31 .34
0.31 0.34
0.31 0.34
0.31 0.34
Compressive strength,
s pl [MPa]
0.04 7.0
1.9 14.0
1.3 1.7
0.9 3.0
0.6 1.1
0.04 7.0
2.0 20
1.6 1.8
0.9 2.7
0.6 1.1
Tensile strength,
s UTS [MPa]
0.05 8.5
2.2 30
1.6 1.9
1.9 3.5
1.0 2.4
Endurance limit,
s ce [MPa]
0.02 3.6
0.95 13
0.9 1.0
0.45 1.5
0.3 0.6
Densification strain, eD
0.6 0.9
0.4 0.8
0.7 0.82
0.8 0.9
0.9 0.94
0.01 0.02
0.002 0.04
0.01 0.06
0.1 0.2
0.03 0.1
Fracture toughness,
KcIC [MPa.m1/2]
0.03 0.5
0.3 1.6
0.1 0.9
0.1 0.2
0.6 1.0
Thermal conductivity,
l [W/mpK]
0.3 10
3.0 35
3.5 4.5
6.0 11
0.2 0.3
90 3000
20 200
210 250
180 450
300 500
The parameters that influence the structure-sensitive properties of cellular metals are (ordered by their importance):
intrinsic properties (properties of cell wall material),
relative density,
type of cellular structure (open or closed cells),
in a closed-cell foam, the fraction of the solid contained in the cell nodes, edges
or the cell faces,
irregularity or gradients in mass distribution,
the cell size and size distribution (including exceptional sizes),
shape of the cells and the anisotropy of cells (including exceptional shapes),
connectivity of cell edges, and
defects, by which we mean buckled or broken cell walls.
x
x
x
x
x
x
x
x
x
5 Material Properties
Some scaling relations have been developed for idealized structures [2], between
the properties of the solid, the density, and the type of cellular structure: a few
of them are listed here. These relations clearly point out that the most important
structural characteristic of a cellular material is the relative density r/rs.
The importance of other structural parameters can be easily seen in the density
dependence of the thermal and electrical conductivity. The thermal conductivity of
metal foam is mainly dominated by the conductivity through the solid. The contribution of the gas, radiation across the pores, and convection within the cell play a
minor role. Therefore, the thermal and the electrical conductivity should behave in
the same manner. The conductivity of the cellular metal should be equal to the
conductivity of the dense solid times its volume fraction (r/rs) multiplied by an efficiency factor, which is given by the geometry and takes into account the tortuous
path of the heat flow. If we assume that a variation of the relative density is caused
by a proportional change of the thickness of the cell wall and struts, or an increase
of the size of the cells maintaining the cell wall thickness the efficiency factor
should be constant. For Al foams it was shown that a variation of the efficiency factor can be expressed also in a power law of the relative density, it is about proportional (r/rs)0.5, hence the conductivity is about proportional (r/rs)1.5.
In contrast to thermal and electrical properties, the influence of the density and
the architecture of the cellular metal on the mechanical properties is much stronger and more complex. For load bearing structures, packaging, and crash elements
their importance is evident, however, for functional applications, for example, heatexchange systems, damping, and filtering the structural stability is also essential.
In general, one can distinguish between elastomeric, elastoplastic, and brittle
foams (this is not a classification of the solid material, it describes the behavior
of the cellular structure).
In compression, these three types of foam show a similar behavior [2] but for
different reasons. Metallic foams, having a relative densities of r/rsI0.2, exhibit
after an elastic loading a more or less clear plateau regime, leading into a final regime of steeply rising stress. The differences are visible in the mechanism of deformation or by unloading of the compressed cell structure: in elastomeric
foams the collapse is by elastic buckling of the cell walls and is more or less recoverable; in plastic foams it is the formation of plastic hinges; and in brittle foams it
is the brittle fracture of the cell wall (crush) that causes the plateau region: the last
two are not recoverable.
Cellular metals are usually ductile. Figure 5-1 compares schematically the
stress strain curve of a cellular and a solid metal in tension and compression.
The foam shows a linear elastic regime, which is not so clearly established as in
a solid; followed by a hardening, which is not only hardening as in a solid, it
is induced by a redistribution of deformed regions; in compression one reaches
then a plateau regime and a final densification; in tension the maximum stress
is reached at relative small strains, typically 1 4 %, which is much smaller than
in a solid metal. The characteristic values are the Young's modulus E (strictly
the stiffness of the cellular structure), the elastic limit s y (which may be slightly
different in tension and compression), the tensile strength, s UTS, the plateau stress
181
182
5 Material Properties
Figure 5-1.
Schematic comparison of the stress strain behavior of a metal and a metal foam.
5 Material Properties
References
5.1
5.1.1
Young's Modulus
One of the most important properties of structural materials is the linear elastic
behavior, which is described by a set of moduli. For isotropic materials two moduli
are needed: the Young's modulus, E, and the shear modulus, G, to characterize the
linear elastic response. Some foams are not isotropic owing to the manufacturing
processes and therefore more than two moduli are necessary to describe the linear
elastic behavior.
183
184
In the isotropic case, which will be accounted for in this section owing to simplification and because many closed-cell metal foams are more, or less isotropic,
the following well known relations between Young's modulus E, shear modulus
G, bulk modulus K, and elastic Poisson's ratio n exist
G
E
E
, K
2(1 S v)
3(1 s 2v)
(1)
5 Material Properties
E
z
Es
r
rs
E
z F2
Es
!2
(open-cell)
r
rs
(2)
!2
S(1 s F)
r
(closed-cell)
rs
The subscript s indicates the properties of the solid from which the cell walls
and cell edges are made. F is the fraction of solid that is contained in the cell edges
for closed-cell foams. Although these equations are based on a simple model of a
cubic foam structure, which considerably deviates from a real foam, there is a fair
agreement with experimental results (Fig. 5.1-1). The equations are valid as long as
the main deformation mechanism is bending of cell edges. For higher-density
foams extension and compression of cell edges becomes more important and
therefore a deviation from these relations should be observed.
Figure 5.1-1. Dependence of the Young's modulus on the relative foam density for different types
of closed-cell (closed symbols) and open-cell (open symbols) aluminum foams [6].
185
186
Influence of Deformation
Deformation changes the structure of a foam by bending, buckling, stretching, and
cracking of the cell edges and cell walls. Therefore, the Young's modulus will
change with strain. In general, there is only a very small linear elastic part in
the stress strain curve of a ductile foam, which makes the measurement of the
initial Young's modulus difficult. For that reason many authors measure an unloading modulus at a certain strain and use this one as Young's modulus, E. It
is important at which strain the modulus is measured because it changes with increasing deformation. Figure 5.1-2 shows the stress strain curves in compression
and tension and the unloading compliance of an Alporas foam (rr 0.1).
It is evident that the Young's modulus with increasing strain |e| decreases much
faster in compression than in tension. This is induced by the fact that buckling cell
edges and cell walls reduce the stiffness much stronger whereas stretching will
increase the stiffness before the initial cracking of cell walls during tension loading. Ultrasonic test methods apply a very small deformation on the foam and permits the measurement of the Young's modulus during deformation experiments.
The obtained Young's moduli are significantly higher than those measured in standard compression or tension tests [7]. Figure 5.1-3 shows the dependence of the
Young's modulus on the deformation obtained with a ultrasonic test method.
For structural applications usually an initial Young's modulus is needed. Since
the determination of the initial Young's modulus is difficult, it is suggested to
use the unloading modulus E0.2 at 0.2 % total strain in compression, or tension,
which seems to be a good compromise between measurability and the effect of
plastic deformation.
5.1.1.3
Figure 5.1-2. Stress strain response with measured unloading moduli indicated for a closed-cell
aluminum foam (Alporas, rr 0.1).
5 Material Properties
Figure 5.1-3. Dependence of the
Young's modulus (stiffness) on
the deformation at which the
modulus is measured [7]. The
change of density is used as
measure at the deformation in
compression.
5.1.2
Compression Behavior
In compression, cellular metals show a unique stress strain response with a plateau region in which the stress is nearly constant over a wide range of strain
(Fig. 5.1-4). This behavior makes cellular metals interesting for energy absorbing
applications where at a relatively low constant stress a large amount of deformation
can be absorbed. More about energy absorbing and crush behavior can be found in
Section 5.1.3. Depending on the material from which the foam is made, different
deformation mechanisms (elastomeric, brittle, and ductile behavior) can be observed. Metallic foams usually show a considerable ductile behavior.
Many investigations have been performed on compression behavior and energy
absorbing performance of metal foams in recent years. The behavior is similar to
many polymer foams. Thus the basic deformation mechanisms and the stress
strain response are well known. Compression loading at small strains of ductile
foams yields to bending and extension/compression of cell edges and cell walls.
If the stresses in the edges and walls exceed the yield stress s y,s of the solid, the
onset of plastification is reached and the deformation is no longer reversible.
Caused by the inhomogeneous structure of real cellular metals the stress concentrations exceed the yield stress in some elements at relatively low strains, which
leads to an early plastification of the foam. Therefore the linear elastic part of
the stress strain curve of a ductile foam is in general hard developed. Increasing
the load on the foam causes to buckling of cell edges and walls in weaker regions
of the foam. A deformation band perpendicular to the loading direction develops,
in which plastic collapse of the cells takes place. This is accompanied by the beginning of the plateau region in the stress strain curve. With increasing strain, additional deformation bands are formed until most of the cells have collapsed and the
187
188
Stress strain response in compression for a ductile aluminum foam and a brittle
alumina foam [11].
Figure 5.1-4.
densification region is reached [8,9]. Depending on the cellular structure and the
properties of the solid, the plateau region of the stress strain curve may not be
flat, but a slight slope, or waviness of the curve can occur. Two extreme cases
are depicted in Fig. 5.1-4, the behavior of a ductile aluminum foam and a brittle
alumina foam, showing a smooth and a jagged plateau, respectively. For the estimation of the plateau stress s pl depending on the foam density r and the yield
stress s y,s of the solid, Gibson and Ashby used a simple cubic-cell model for the
foam. The plateau stress is reached, when the moment exerted to the cell edges
exceeds their fully plastic moment
s pl
z 0:3
s y; s
s pl
z 0:3
s y; s
r
rs
F
!3=2
(open-cell)
r
rs
(3)
!3=2
S0:4(1 s F)
r
(closed-cell)
rs
Figure 5.1-5 shows the stress strain curves in compression of Alulight aluminum foams with different densities.
The increase of the plateau stress with increasing density is clearly obvious. Different methods exist for the measurement of the plateau stress depending on the
course of the stress strain curve. If there is a stress peak at the beginning of the
plateau, this peak or a clear flat section stress will be taken as plateau stress [10]. In
5 Material Properties
Figure 5.1-5.
compression.
the absence of a stress peak the plateau region will be extrapolated to lower strains
and the intersection with the initial elastic line, or the ordinate (Rpe) delivers the
plateau stress [10,27]. The plateau stress plays an important role to characterize
the energy absorbing behavior and is a good material property for the compression
performance of a foam. However, for structural design with foams the plateau
stress may not be applicable, because of the relatively high deformation and
damage of the foam that is accompanied with this stress level. The determination
of a yield stress s y is problematic owing to the usually poorly developed linear
elastic regime of the stress strain curve. To compromise, an offset yield stress,
like s 0.2, or Rp02 should be used, depending on the mechanical response of the
cellular metal.
For the experimental determination of a stress strain curve, standard equipment for compression tests can be used. If the measurement of the Young's
modulus E and the 0.2 % offset yield stress s 0.2 is needed, the displacement should
be measured directly at the specimen with a clip gauge, or a videoextensiometer. To
avoid errors arising from inhomogeneous deformation of the foam the displacement should be measured at more than one position on the specimen. The
specimen size must exceed the average cell size by 7 to 10 times in each principal
direction to reduce the influence of the inhomogeneous cell structure.
189
190
5.1.3
Good energy absorbers should support large strains at a relatively low constant
stress level. Owing to their compression behavior, cellular metals (especially
foams) and also some polymer foams, are especially suited for energy absorbing
applications. The advantages of metal foams for these applications are their relatively low density, adjustable plateau stress level, good temperature resistance
and non-flammability. Thus their application fields are the transport and packaging
industries.
Figure 5.1-6. Comparison of stress strain re- the same for both. The peak stress of the absponse of two different energy absorbers. The sorber that has no plateau region is higher and
amount of absorbed energy per volume in a
this will enhance the impact forces.
certain strain interval is indicated as EV and is
5 Material Properties
Figure 5.1-7. The energy absorbing efficiency, h, is the ratio between absorbed energy of the
material EV and the absorbed energy of an ideal plastic material EV,ideal in a certain strain range.
The absorbed energy per unit volume EV in a certain strain interval [e1, e2] is
equal to the area below the stress strain curve and can be expressed as
Z e2
s(e)de
EV
(4)
e1
The efficiency of energy absorption, h, is the ratio between the absorbed energy of
the real material and the absorbed energy of an ideal absorber that is an ideal plastic material
R e2
s(e)de
(5)
h e1
s 0 (e2 s e1 )
Most cellular metals have a nearly flat plateau region and therefore a high energyabsorbing efficiency depending on the used strain interval, and in general, a good
absorbing capacity (Fig. 5.1-7).
Impact Behavior
For the application of cellular metals as crash absorbers in transport systems not
only is the energy absorbing capacity but also the impact behavior important. At
high deformation rates, which may occur in crash situations, the mechanical
properties of the foam may change. There are two points that must be taken
into consideration: the inertial mass of the cell edges and walls and the gas pres5.1.3.2
191
192
sure and gas flow inside the cells of a closed-cell foam. Both effects can increase
the flow stress and therefore also the plateau stress of the foam with increasing
strain rate [12]. In open-cell structures no significant increase of flow stress is observed at higher strain rates, which suggests that the inertial masses of the structural elements have no contribution to the flow stress [12]. In closed-cell foams the
situation is different, because some of the metal foams show a strain-rate dependence. Real closed-cell metal foams are, in general, not completely closed-cell structures because they frequently contain some initial defects in the cell walls arising
from the manufacturing process. Additional defects are generated during the deformation process when cell walls rupture. The gas flow between the cells through
these defects is strain rate dependent. Thus the gas pressure inside the cells increases during high-speed deformation and contributes to the flow stress [13].
Owing to the strain localization in deformation bands, the local strain rate is
much higher than the measured global one, which will increase the strain rate
effect. As a result, the plateau stress and the energy absorbing capacity of closed
cell metals foams ise [12,13]. Investigations on aluminum foams show an increase
in the plateau stress of 20 90 % when the strain rate increases from 10 3 to 103 s 1
(Fig. 5.1-8). The strain rate effect is mainly controlled by the cell morphology and
the alloy from which the foam is made.
Figure 5.1-8.
Strain-rate dependence of the plastic strength of two types of aluminum foams [12].
5 Material Properties
5.1.4
Tension Behavior
p
s cr
z C l=a
s fs; s
r
rs
!3=2
(open-cell)
(6)
where C is a constant, l is the cell size, a is the critical initial defect size, and s fs,s is
the fracture strength of the solid. For closed-cell metals the situation is more complicated.
In ductile cellular metals (such as aluminum foams) cracking of cell walls and
cell edges can also be observed before a main crack starts to propagate. With increasing deformation, a fracture-process zone starts to develop within which the
whole deformation is concentrated. In closed-cell metals, several cell walls rupture
inside this fracture-process zone. Further deformation leads to development and
propagation of a main crack along the weakest path in the cellular structure.
Only the strongest cell edges remain intact and hold the two foam parts together.
This process is accompanied by a decrease in flow stress, which is evident in the
stress strain curve depicted in Fig. 5.1-10. Figure 5.1-9 shows the corresponding
193
194
Failure mechanism in aluminum foam in tension. Crack initiation in some weak cell
walls and subsequent development of a fracture-process zone (stress strain curve in Fig. 5.1-10).
Figure 5.1-9.
Figure 5.1-10. Stress strain response in tension of aluminum foam with some unloading moduli
indicated (Alporas, 3r 0.18).
5 Material Properties
Figure 5.1-11. Influence of the notch depth ratio, a/W, on the stress strain response of a ductile
aluminum foam in tension.
195
196
5.1.5
Torsion Behavior
In torsion the cell edges are subjected to bending and the cell walls to a combination of bending and shearing. The shear behavior is important for the development of material laws and also for the application in sandwich structures, where
core shear may occur in particular load situations. Alternative methods for testing
the shear strength are, for example, the double-lap shear test or the ASTM C-273
Stress strain response of aluminum foam in torsion with no axial stress (black
line) and with tensile axial stress (gray line) [17].
Figure 5.1-12.
5 Material Properties
Fracture Behavior
Fracture behavior and fracture toughness values may play a major role for the application of cellular metals foams in load-bearing structural components. The fracture of brittle foams, where linear elastic fracture mechanic (LEFM) can be applied,
were discussed briefly in the section on tensile behavior.
The situation in closed-cell ductile metal foams is more complicated, where
LEFM usually can not be applied, which is considered in this section.
197
198
Figure 5.1-13. Diagram of the load displacement curve and the corresponding SEM images to
demonstrate crack initiation and crack propagation in a ductile aluminum foam acquired from insitu fracture tests.
Fracture Toughness
For ductile cellular metal foams, linear elastic fracture mechanics is usually not applicable, because they show an extensive plastic regime and the specimen sizes are
limited by the manufacturing processes. Therefore elastoplastic fracture mechanic
tests, like J-integral tests, or COD (crack opening displacement) tests, must be used
to determine the fracture toughness of metal foams. Common standards, like
ASTM E813 for J-integral tests, are optimized for ductile solids and can only be
used in a limited manner for foams. Size restrictions for the specimen geometry
5.1.6.2
5 Material Properties
to obtain valid fracture toughness values should be thought about because, for
example, the required specimen size also depends on the cell size, l, of the foam.
For the determination of J Da curves according to ASTM E813 or ASTM E1152,
the following parameters must be measured: the force F, the load-line displacement vLLD, and the crack extension Da. Figure 5.1-14 shows a load versus loadline displacement curve obtained in a fracture mechanics test and the crack extension as a function of the load-line displacement of a ductile aluminum foam. The
calculated J Da curve is plotted in Fig. 5.1-15. The ASTM standard provides that
the initial J-value, J0.2, is obtained by the intersection of the blunting line at
0.2 mm crack extension with the straight line achieved from a linear regression
of the J Da curve in the Da interval 0.15 1.5 mm.
A crack extension of 0.2 mm is rather low for metal foams, because it is one
magnitude smaller than the usual cell size (the characteristic microstructural
length) and can be caused by other effects (microcracking, localized yielding)
rather than crack extension, depending on the measurement method used. Fleck
and Ashby [19] introduced a steady-state J-value, JSS, which is measured as the plateau J in the J Da curve. They assumed that this JSS is a measure of the fracture
toughness for cracks, where the bridging zone is fully developed. However, not all
cellular metals show a typical plateau region in the J Da curve and the plateau is
reached at rather high crack extensions. The standardization of the evaluation of
initial, or steady state fracture toughness values from J Da curves is a problem
that has not yet been solved satisfactorily.
Gibson and Ashby [6] established the following relation between the fracture
toughness and the foam density for ductile metal foams
Figure 5.1-14. Load and crack extension versus load-line displacement curve for a ductile metal
foam (Alporas, rr 0.1).
199
200
J-integral versus crack extension plot with fracture toughness values JSS and J0.2
indicated (J-values obtained from Fig. 5.1-14).
Figure 5.1-15.
JIC z Cs y; s l
r
rs
!3=2
(7)
This equation correlates the fracture toughness with the cell size (compare with
Eq. 8, it is assumed that COD l), which is demonstrated only for a limited number of foams and densities. Other fracture toughness values, like KIC, or CODC, can
be derived from this equation using the well-known fracture mechanics relations.
An alternative method for the determination of the fracture toughness of a ductile material is the measurement of the critical crack opening displacement (COD).
Although this is a direct method, it is rarely used owing to the difficulties in
experimental determination. Schwalbe [20] suggested to measure the COD 5 mm
away from the crack tip, COD5, which can be more easily performed. We have
adapted this technique to foams. Figure 5.1-16 shows COD5 Da and CTOD Da
curves of a ductile aluminum foam (CTOD crack tip opening displacement).
The advantage of this method is the similarity of the COD Da curves for different
foams and the easier evaluation of critical fracture toughness values. The relation
between COD and the J-integral is given by the following equation
J k s COD
(8)
where s * is the mean value between yield stress s y and ultimate tensile stress s UTS
and k is 1 for plain stress state and 1,3...2,6 for plain strain state for common
metals. In the case of metal foams the value of k for the (macro) plain strain
state depends on the Poisson's ratio of the foam in the plastic regime. It is im-
5 Material Properties
portant to note that we have to distinguish the macro- and micro- stress states (the
stress state in the cell wall). Since npl is much smaller than in solid metals, k should
be only somewhat larger than 1.
201
202
References
5 Material Properties
5.2
203
204
quently, the load ratio R is 1 for fully reversed loading, 0 J R I 1 for cyclic tension loading and 1 J R J T for cyclic compression loading. The amplitude of a
stress cycle Ds/2 (s max s min)/2 is used to describe the cyclic stress range.
A summary of S N curves described in the literature is shown in Fig. 5.2-1.
The S N data are approximated by power-law curve-fits with a 50 % survival
probability.
Nn
Ds
C
2
(1)
The values of the exponent, n, and the coefficient, C, (in MPa) have been evaluated
and are shown in Table 5.2-1. Different symbols and lines refer to different materials and loading conditions (cyclic compression, cyclic tension, or fully reversed
loading). Fatigue properties of the different aluminum foams show significant variations. The cyclic stress amplitudes that lead to mean lifetimes of 105 cycles, for
Alcan
Duocel
Alporas
Alporas
Alporas
Alporas
Alporas
Alulight
IFAM
Alulight
Alcan
Duocel
Alporas
Alporas
Alporas
Alporas
Alulight
Hydro Al
IFAM
Alulight
#2
#3
#4
#5
#6
#7
#8
#9
#10
#1 1
#12
#13
#14
#15
#16
#17
#18
#19
#20
Foam
[11]
[8]
[8]
[5]
[6]
[2]
[6]
[6]
[2]
[1]
[5]
[3]
[5]
[6]
[2]
[6]
[6]
[4]
[2]
[1]
Ref.
0.56
0.40
0.29
0.86
0.30
0.30
0.25
0.25
0.19
0.15
0.50
0.63
0.45
0.30
0.30
0.25
0.22
0.22
0.19
0.15
Density
[g/cm3]
7.0
4.4
3.5
17.0
2.25
1.9
1.75
1.47
0.53
0.25
19.5
8.8
6.3
2.34
1.9
1.71
1.37
1.4
0.53
0.25
Plateau
stress [MPa]
rever.
rever.
rever.
tension
tension
tension
tension
tension
tension
tension
compr.
compr.
compr.
compr.
compr.
compr.
compr.
compr.
compr.
compr.
Loading
condition
0.1
0.1
0.1
0.1
10
10
10
10
10
10
10
Load
ratio
51
19
19
14
26
33
30
13
30
28
30
Tested
specimens
6.988
15.691
5.296
13.645
2.254
1.355
1.467
1.068
0.554
0.131
12.514
6.232
3.393
1.759
1.332
1.028
0.710
0.626
0.546
0.133
Coefficients
C [MPa]
n
0.110
0.189
0.111
0.090
0.093
0.068
0.080
0.072
0.081
0.108
0.031
0.074
0.038
0.053
0.063
0.038
0.028
0.037
0.079
0.064
List of references for S N curves shown in Fig. 5.2-1. Coefficients C and n refer to a power-law approximation of data (Eq. 1) [1 6,8,11].
#1
Table 5.2-1.
5 Material Properties
205
206
example, may vary by a factor of 200, depending on the investigated foam and the
load ratio.
The main reasons for the large variations of the fatigue properties of aluminum
foams are their different mass densities and structures. Literature data show a significant trend, that fatigue properties of foams increase with increasing plateau
stress. As an example, the stress amplitude for 50 % survival probability of 105 cycles and of 107 cycles, respectively, are shown in Fig. 5.2-2. Irrespective of foam
type and loading condition (cyclic tension, cyclic compression, or fully reversed),
the fatigue strength increases with the plateau stress. The linear relation shown
in Fig. 5.2-2 indicates, that the stress amplitude for mean lifetimes of 105 cycles
is approximately 0.35 of the plateau stress, and Ds/2s pl 0.27 leads to mean lifetimes of approximately 107 cycles. This means that the fatigue lifetime of an aluminum foam is mainly determined by the cyclic stress amplitude, Ds/2 with respect to the plateau stress, s pl, and Ds/2s pl is an appropriate measure of the magnitude of cyclic loading.
To compare different aluminum foams it is reasonable to normalize the cyclic
properties with respect to the plateau stress s pl. Eq. (1) may be rewritten as follows
Nn
Ds
C
2s pl
s pl
(2)
In Fig. 5.2-3, literature data obtained under cyclic compression, under cyclic tension, and under fully reversed loading conditions are summarized. Additionally,
mean S N curves for these three loading conditions are shown. The algebraic expression of the mean curves of the literature data is
5 Material Properties
N nmean
Ds
C
C0
2s pl
s pl
(3)
The values of the exponent, nmean, and the coefficient, C0, of the mean curves are
shown in Table 5.2-2. The largest absolute value of the exponent nmean is found for
fully reversed loading and the smallest for cyclic compression loading.
Using the relative stress amplitude, Ds/2s pl to describe the magnitude of loading, Duocel foam shows good fatigue properties under cyclic compression
Table 5.2-2.
Exponent nmean
0.628
0.0491
0.800
0.0825
1.738
0.1351
a)
b)
Figure 5.2-3.
c)
207
208
(Fig. 5.2-3a: #2) and cyclic tension (Fig. 5.2-3b: #12). Duocel foam was the only
open-cell foam where fatigue data were found in the literature. Similar relative fatigue strength is visible for Alporas foam (Fig. 5.2-3a: #3 #7, Fig. 5.2-3b:
#13 #16), Alulight foam (Fig. 5.2-3a: #8,#10, Fig. 5.2-3b: #17, Fig. 5.2-3c: #20)
and IFAM foam (Fig. 5.2-3a: #9, Fig. 5.2-3c: #19). Hydro aluminum foam has
been tested under fully reversed loading conditions only (Fig. 5.2-3c: #18) and
shows the highest relative strength at R 1. Alcan foam, which is the material
with the lowest density, shows the lowest relative fatigue strength under cyclic
compression and cyclic tension (Fig. 5.2-3a: #1, Fig. 5.2-3b: #11).
5.2.2
To evaluate the high cycle fatigue properties and to determine an eventual fatigue
limit for aluminum foams, specimens have to be tested up to very high numbers of
cycles. These investigations are time consuming using conventional fatigue testing
equipment working at low frequencies, since several specimens have to be tested to
obtain statistically relevant data of this rather inhomogeneous material. Therefore,
the ultrasonic fatigue testing method has been used in this study to investigate the
high cycle regime in an efficient and time saving manner.
5 Material Properties
Results
Results of endurance tests with AlMg1Si0.6 foam are shown in Fig. 5.2-4a. Numbers of cycles to failure are related to the experimentally determined cyclic strain
amplitude, De/2. Assuming a mean Young's modulus of 3.9 GPa, the stress amplitude, Ds/2 can be calculated from the measured cyclic strain as shown in
Fig. 5.2-4a (right ordinate). Specimens that did not fail within 109 cycles (runouts)
are marked with an arrow. Fatigue data are subjected to pronounced scatter, which
is typical for inhomogeneous materials. The solid line indicates a mean fracture
probability of 50 % assuming power-law dependence between stress amplitudes
and cycles to failure.
The S N measurements of AlMg1Si0.6 foam indicate a well-defined fatigue
limit: Specimens either fail before about 107 cycles are reached or they survive
109 cycles or more. The median endurance limit (failure probability of 50 %) is in5.2.2.2
a)
b)
c)
d)
Results of endurance fatigue tests under fully reversed tension compression cycling: a) AlMg1Si0.6 foam, b) AlMg1Si0.6/T5 foam, c) AlMg0.6Si0.3 foam, d) AlSi12 foam.
Figure 5.2-4.
209
210
dicated with a line parallel to the abscissa in Fig. 5.2-4a at a stress amplitude of
1.3 MPa, which is approximately 23 % of the plateau stress. However, specimens
may fail at significantly lower cyclic stresses due to the statistical distribution of
defects and areas of stress concentrations. The highest stress amplitude, where
no specimen failed within 109 cycles, is 1 MPa.
Endurance tests on AlMg1Si0.6 T5 foam are summarized in Fig. 5.2-4b. The
heat treatment leads to longer mean lifetimes at all stress levels in comparison
with the not heat treated material. Additionally, the endurance limit increased.
The maximum stress amplitude, where no specimen failed, is 1.2 MPa, and the
median fatigue strength at 109 cycles is 1.4 MPa (approximately 20 % of the plateau
stress).
The results of endurance tests with AlMg0.6Si0.3 and AlSi12 foams are shown in
Fig. 5.2-4c and d, respectively. At a stress amplitude of 1 MPa, one (of five tested)
AlMg0.6Si0.3 foam specimens failed. The fatigue limit for 50 % survival is 1.1 MPa
(approximately 22 % of the plateau stress). The median fatigue strength at 109 cycles of AlSi12 foam is 1.3 MPa (approximately 19 % of the plateau stress). Fatigue
data of AlSi12 foam are subjected to the most pronounced scatter among the investigated Alulight foams. At stress amplitudes of 1.2 MPa three specimens did not
fail within 109 cycles, whereas one specimen failed at 1 MPa.
5.2.3
5 Material Properties
Figure 5.2-5.
cooling.
SEM studies at higher magnification show, that the fatigue fracture surface is relatively smooth and transcrystalline. The general fracture appearance is ductile,
and striations are visible in some areas. Figure 5.2-7 shows both, a fatigue fracture
surface in a cell wall (Fig. 5.2-7a and b) and a fracture surface that has been formed
by static rupture of the specimen after fatigue testing (Fig. 5.2-7c). Fatigue cracks
show clear differences compared with cracks formed by static loading or during
cooling of the melt due to the far lower plastic deformation of the material around
the crack path. Fatigue crack paths preferentially follow areas where the cell wall is
thin. In Fig. 5.2-7, the thicknesses of cracked cell walls are approximately
30 60 mm, whereas the most frequent wall thickness obtained by image-analysis
is 250 500 mm [10]. SEM studies of the fracture surface do not show deformation
bands or plastic buckling as it could be found for uniaxial compression or for cyclic
compression fatigue loading.
The role of cell nodes, where fatigue cracks may be trapped, helps to understand
the influence of different foam structures on the fatigue properties. Duocel is an
211
212
Typical fatigue cracks: a) fatigue crack inside a cell wall, length about 300 mm,
b) crack starting from a cell-wall defect, length about 150 mm, c) fatigue crack causing rupture
of a specimen.
Figure 5.2-6.
5 Material Properties
Figure 5.2-7. Fracture surfaces of AlMg1Si0.6 T5 (Ds/2 1.96 MPa, number of cycles to failure
8.3 q 104): a) fatigue fracture surface, b) striations indicating crack growth direction, c) fracture
surface formed by static rupture.
213
214
open-cell foam with r 0.15 g/cm3 and consists of cell nodes only and approximately the same amount of material is distributed to nodes and cell walls inside
Alcan foam (r 0.19 g/cm3). Fatigue cracks initiated in Alcan foam may find
lower growth resistance since cell nodes contain less material compared with Duocel foam. This may be one reason, why Duocel foam (Fig. 5.2-3b, #11) shows far
better cyclic tension properties than Alcan foam (Fig. 5.2-3b, #12).
5.2.4
Summary
A survey of literature data (20 S N curves determined with six different aluminum
foams under cyclic compression, cyclic tension, and fully reversed loading) shows
that the fatigue strength is related to the plateau stress, s pl.
x
The stress amplitude, Ds/2 for mean lifetimes of 105 cycles (107 cycles) is approximately 35 % (27 %) of the plateau stress.
Numbers of cycles to failure and relative stress amplitudes, Ds/2s pl may be correlated by a power law function. The largest gradient is found for fully reversed
loading and the smallest for cyclic compression loading.
To investigate the high cycle fatigue properties of aluminum foams, three Alulight
foams have been investigated at numbers of cycles between 104 and 109. Cylindrical
rods with a continuous surface layer and a closed cell structure were tested. The
foam density is 0.56 g/cm3.
x
The investigated Alulight foams show a fatigue limit, and cycles to failure above
107 are rare. The median endurance limits, Ds/2 are 1.1 1.4 MPa, which is
19 23 % of the plateau stress.
Fatigue damage under fully reversed tension-compression loading is governed by
the formation of cracks. Preferential areas for crack initiation are initial defects
like precracks or holes in the interior sections of cell walls. No strain localization
or formation of deformation bands (as reported for cyclic compression tests) is
found.
Fatigue cracks preferentially grow in the inner sections of cell walls where wall
thickness is small. Below the endurance limit eventual cracks are trapped at
the nodes of cells.
Acknowledgements
The authors acknowledge financial support by the Fonds zur Forderung der wissenschaftlichen Forschung (FWF), Vienna, (project no. P13230PHY) and the LKRCenter of Competence on Light Metals, Ranshofen, Austria, for the grant through
the K-plus project and for supplying the Alulight foams.
5 Material Properties
References
5.3
It is evident that the presence of gas bubbles will significantly affect also electrical,
thermal, and acoustic properties of cellular solids. In the case of polymer foams
these properties play an important role: thermal insulation and noise attenuation
are beside packaging the main fields of application for this kind of materials. However, only a little attention is paid to the mentioned properties when cellular metals
are considered, although the multifunctionality is often the prerequisite for their
successful application. The aim of this chapter is therefore to reveal and critically
discuss the potential of cellular metals in this field with respect to published models and rarely reported experimental data. For better understanding of considered
properties, the basic physical background will be also briefly introduced.
5.3.1
Electrical Properties
Cellular metals are in distinction to polymer and most ceramic matrices electrically
conductive. This simple fact changes the typical applications drastically: ceramic
and polymeric foams can be used for insulation or for structural enclosures that
215
216
must transmit microwaves, while the metallic counterparts are good for the opposite purpose. The electrical conductivity of foamed metals, although considerably
reduced by the porosity, is still sufficient enough to provide good electrical grounding, low voltage contacts, or capability to absorb electromagnetic waves. In the case
of metallic sponges with open cell structure, the conductivity allows to distribute
the electric potential on large accessible area, which makes them attractive for battery electrodes. Cellular nickel is extensively used in this application [1].
It is generally accepted, that cellular material inherits electrical characteristics of
the solid of which it is made. The electrical conductivity s of material is defined
according to Ohm's law
s
l I
S U
(1)
where U is a potential drop along the distance l when current I is applied. As can
be seen conductivity depends strongly on the cross-sectional area S of the sample.
In a case of cellular metals only a small portion of cross section is electrically conductive. These are the cell walls with roughly the conductivity of the base alloy. The
main part of the cross section is created by pores, which are usually filled with dielectric air. The cell walls form a complex 3D network of electric conductors of
varying cross section, whereas cell walls oriented normally to the potential gradient
do not contribute to conduction because of no potential difference. Moreover, the
cells are mostly continuously covered with electrically nonconductive metal oxides.
The cellular structure provides therefore much lower electrical conductivity than
bulk material.
To calculate the electrical conductivity according to Eq. (1) the effective length l of
the cell walls conducting along the potential difference and in the cross-sectional
area S effectively contributing to conduction has to be known. Unfortunately, it
is extremely difficult to determine it in real samples, although for example computed tomography offers this possibility. More reasonable and cheaper is to consider the sample as a black box of given outer geometry with macroscopically
homogeneous and continuos distribution of a material within the box and to investigate the so called effective electrical conductivity.
It is evident that the dependence of effective cross-sectional area on relative
density of the structure is non linear. One of the first models of this dependence
was introduced by Ashby et al. for the case of open-cell sponge [2]. The structure
was idealized with narrow struts of given length and thickness. The struts were
arranged in a perfect cubic lattice, with nodes linked by struts. Only struts oriented
parallel to potential gradient contributed to conduction. For higher relative density they added also the contribution of the nodes and obtained the following
relation
s
r
f
s0
r0
!
S(1 s f)
r
r0
!3=2
(2)
5 Material Properties
where s and r are the effective property and density of the foam respectively, while
s 0 and r0 are the corresponding properties of the cell-wall material. The first term
of this equation describes the contribution of struts and the second one that of
nodes. Therefore, a structural parameter f was introduced, which characterizes
how the material is distributed between cell nodes and struts. However, in real
sponges the ratio between volume of cell nodes and cell struts is not constant
and varies also with the density. Thus f in Eq. (2) must be obtained experimentally
for each material and is valid only for certain density interval.
Another approach for modeling of the relationship between relative density and
selected physical property uses the principles of percolation theory [3 6] and
covers the whole density range. In this case the cellular metal is considered as
an infinite cluster of randomly distributed gas pores in the metallic matrix.
According to percolation theory the effective property near the connectivity
threshold (in this case set to zero) behaves like a power-law function of a relative
density
!t
s
r
z
(3)
s0
r0
where s and r are the effective property and density of the foam respectively, while
s 0 and r0 are the corresponding properties of the cell-wall material. Constant t
(often called as a critical exponent) can be theoretically predicted for the specific
property. The theoretically estimated value for electrical conductivity in 3D is t
2.00 [7]. The constant z ought to be 1, because for bulk material r r0 and the
effective property s s 0.
Also this approach has some difficulties: It does not take into account defects
such as cracks in pore walls, which influence the conductivity but not the relative
density. Moreover, the theoretical predictions of critical exponents in 3D are still
under development and optimization, especially when finite size clusters or anisotropic orientations are considered.
Only little data for electrical conductivity of cellular metals have been reported
[8 12] up to now (see also Table 5.3-1). Therefore, a serious verification of recent
theoretical models is very difficult, if possible at all.
f 0.33 and f 0.05 were obtained for ERG open-cell sponge and for Alulight
closed-cell foam, respectively, applying Eq. (2) to fit the reported data [2]. The considerable difference in a, confirms an assumption that a is strongly material and
structure dependent and for each case has to be obtained experimentally.
The power law seems to be a good characterization of the relationship between
experimental data of electrical conductivity and density measured on Alulight samples [3 5] (see Fig. 5.3-1 and Table 5.3-2). The characteristic exponents obtained
experimentally have been found slightly lower than the critical exponent value theoretically predicted by the percolation theory. It can be explained by the fact that in
distinction to the theory, where an infinite cluster is taken into account, the real
experimental samples have always a finite size and in the referred case possess
a surface skin. The volume fraction of the surface skin increases with decreasing
217
218
Table 5.3-1.
Suppliers
Density [kg.m 3]
Thermal conductivity
[W/mK]
Electrical conductivity
[m/Vmm2]
Alporas [10]
570
9.9
400
5.0
300
3.8
200
2.6
94
1.0
500
13
3.1
400
10
2.6
300
8.9
1.9
540
0.91
410
0.7
270
0.4
160
0.36
68
0.28
330
2.1
250
1.6
200
1.2
140
0.9
300
2.1
0.50
250
2.1
0.36
150
1.5
0.27
Alulight [9]
Cymat [11]
ERG [8]
size of the sample. The high portion of surface skin can diminish the effect of the
porosity on the reduction of electrical conductivity of the foam. If the surface skin
is removed from the sample the reduction of the conductivity due to increasing
porosity is enhanced and characteristic exponent becomes higher. The similar effect can be observed when the sample size increases. It implies that the conductivity of the foam is not only a function of the porosity, but it is strongly affected also
by the shape and size of the foamed part.
5 Material Properties
Table 5.3-2. The obtained fitting parameters for the dependence of normalized electrical
conductivity on the relative density for Alulight Al 99.7. x 2 is a minimization function [7].
x2
0.999 e 0.009
1.55 e 0.02
7.31 qx 10
0.979 e 0.025
1.75 e 0.06
6.54 qx 10
Although the behavior can be qualitatively understood by noting that with decreasing density the cross section available for conduction decreases and the tortuosity of the current path increases, a quantitative theory is still lacking [13].
Good electrical conductivity of cellular metals minimizes the penetration of electromagnetic waves into the structure, especially in the case of closed-cell foam
parts covered by a surface skin. This fact can be successfully utilized for the protection of electronic devices or humans from the effect of electromagnetic noise.
The ability for electromagnetic wave shielding can be defined by shield effectiveness depending on the wave frequency and the material thickness. The main
advantage of metallic foam is a possibility to achieve a thickness required for
low magnetic permeability at lower weight than in the case of bulk metals. The
experimental measurements on closed cell Alulight and Si-steel sheet of the
same weight are illustrated in Fig. 5.3-2 and Fig. 5.3-3 [9]. The shielding effectivity
of the foam is even slightly better than that of a bulk aluminum plate of the same
thickness, additionally the foam weight was approximately five times lower in this
case (see Fig. 5.3-2). Cellular aluminum foam exhibits much better magnetic field
Figure 5.3-1. The power-law dependence of normalized electrical conductivity on the relative
density for Alulight Al 99.7 foam (cylindrical samples with surface skin, s 0 37.6 q 106 S m 1,
r0 2700 kg m 3) [4], for the fit parameter see Table 5.3-2.
219
220
Figure 5.3-2. Magnetic field shielding effectiveness as a function of frequency for cellular
aluminum (Alulight) and Si-steel samples of the same weight and a massive Al sheet of the same
thickness (t 8.5 mm) as the aluminum foam. Sample size 140 q 140 q t, density of foam
500 kg m 3 (KEC-method) [9].
Electric field shielding effectiveness as a function of frequency for cellular aluminum (Alulight) and Si-steel samples of the same weight. Sample size 140 q 140 q t, the density
of foam 500 kg m 3 with t 8.5 mm (KEC-method) [9].
Figure 5.3-3.
shielding properties than Si-steel sheet of the same weight. The electrical shielding
effectivity of Alulight is comparable with the properties of Si steel (see Fig. 5.3-3)
up to 10 MHz. For higher frequencies the aluminum foam is superior. Similar
results have been reported for Alporas foam [10]
5 Material Properties
5.3.2
Thermal Properties
The main thermal properties of practical importance with respect to cellular metals
are the melting point, specific heat, coefficient of thermal expansion, thermal conductivity, thermal diffusivity, emissivity of the surface, fire and thermal shock resistance. Some of these properties are almost the same as for bulk materials, other
ones depend strongly on the porous structure.
Melting point of cellular metals is practically the same as for an alloy from which
it is made. However, the surface is often covered by a continuous oxide layer, the
melting point of which is usually considerably higher: for example in the case of
cellular aluminum. The surface area covered with oxide layer increases with increasing porosity and decreasing pore size. When the oxide layer is sufficiently
thick it can support the porous structure even above the melting point of the
alloy, provided that no significant external force is applied on the structure. If
the structure is heated in air or oxidizing atmosphere at the temperature close
to melting point for sufficiently long time, the thickness of oxide layer increases
and material becomes more stable. After a threshold thickness depending on the
mechanical stability of the sample, the metallic cell walls oxidize completely by
subsequent heating converting the cellular metal into a ceramic foam. This behavior is utilized for instance for the preparation of ceramic sponges from open-cell
metallic precursors like Duocell sponges [14].
The energy needed to increase the temperature of cellular structure by a unit
temperature is almost the same as required for cell-wall materials when the unit
mass is considered. The slight difference can be accounted to the presence of
thin surface oxides and the contained air. However, the specific heat per unit volume
Cv of a cellular structures is of cause significantly lower. It makes them attractive
for applications where low thermal capacity is required, for example for rapid heating and cooling systems.
The coefficient of thermal expansion a of cellular metals is also almost the same as
for the cell-wall material, while the thermal conductivity l is much lower. Higher
thermally induced distortion has to be expected that increases with raising a/l ratio.
To avoid it, the temperature differences in the cellular structure on heating or cooling should be minimized.
On the another hand if the temperature shock resistance is considered, the situation is not so unambiguous. When a sudden temperature difference occurs between outer surface and inner material regions the thermal expansion causes thermally induced strains, the related elastic stress depend on the modulus of elasticity
of the material. The stresses depend on the product a E and as the modulus of elasticity of cellular metal decreases with decreasing relative density the same temperature difference induces smaller thermal stresses the higher the porosity of the cellular structure, assuming the coefficients of thermal expansion is constant. However, with decreasing density also the collapse or tensile stress of the material decreases. If this stress scales with relative density at smaller exponent than the modulus of elasticity does, the temperature shock resistance of the foam will improve.
221
222
If not, the foam may yield or even break earlier than bulk material. In distinction to
bulk materials, cellular metals are able to withstand much higher compressive
strains without fracture because of the existence of a plateau stress. Therefore
their temperature shock resistance is usually considered to be higher than that
of bulk metals. It should be noted, that the same sudden change of ambient temperature causes higher temperature difference within a foam than within solid
metal because of significantly smaller thermal conductivity of the cellular metal.
It is evident that the thermal conductivity, l, of cellular metal will be significantly
lower than the corresponding conductivity of bulk material. In comparison with
electrical conductivity this property is more complex. Apparent thermal conductivity of porous structure has four contributions [13]: besides the thermal conductivity
of the solid cell walls also the conductivity of enclosed gas, convection and radiation
effects should be taken into account
l lS S lG S lc S lr
(4)
5 Material Properties
Table 5.3-3. The obtained fitting parameters for the dependence of normalized thermal conductivity on the relative density for Alulight foam at various temperature. x 2 is a minimization function.
x2
1.000 e 0.006
1.60 e 0.03
2.97 qx 10
1.000 e 0.008
1.55 e 0.03
5.66 qx 10
1.000 e 0.010
1.48 e 0.04
9.10 qx 10
1.000 e 0.005
1.51 e 0.03
4.00 qx 10
As the thermal conductivity of bulk metals increases with increasing temperature, the same behavior is expected also in a case of cellular metals, in accordance
with Eq. (5). Experimental values determined for Alulight foams made from various aluminum alloys at higher temperatures are shown in Fig. 5.3-5. As expected,
with increasing temperature the effective thermal conductivity of foam increases.
However, a little drop of this property has been observed at temperatures above
300 hC. This small reduction may be caused by heat losses due to increasing radiation effects, but more probably by extensive oxidation of the foam surface at higher
temperatures. The alloys (such as AlSi12), which are less sensitive to oxidation,
exhibit also less reduction of the effective thermal conductivity at higher temperatures.
Figure 5.3-4. The power-law dependence of normalized thermal conductivity on the relative
density for Alulight foam at 20 hC. The corresponding fit parameters are given in Table 5.3-3
(cylindrical samples with surface skin, l0 232 W m 1 K 1, r0 2700 kg m 3) [3].
223
224
Thermal diffusivity a governs heat flow through a material before a steady state
flow by thermal conduction is achieved. It is defined as a ratio l/Cv between thermal conductivity l and volumetric specific heat Cv. Foams have usually higher thermal diffusivity than bulk cell-wall materials and thus they achieve a steady state heat
flow earlier. This property makes the cellular metals attractive for use as heat exchangers.
The emissivity of the foam surface is almost always higher than that one of the
metal from which it is made of. As the actual surface area of the foam is also
much larger at equal outer dimensions the foam will emit or absorb more thermal
radiation than bulk material. This property in combination with small specific heat
per unit volume can be utilized in applications where rapid heating or cooling by
radiation is required.
Although polymer and ceramic foams are used more for thermal insulation than
for any other purpose, application of cellular metals in this field is unlike because
their thermal conductivity l is about two orders of magnitude higher than polymers (see Table 5.3-1). However, higher thermal conductivity and diffusivity of cellular metals can be successfully utilized in heat exchangers or heat sinks for power
electronics, air-cooled condenser towers and regenerators. This application is restricted only to open cell structures enabling flow of cooling gas or fluid without
significant pressure drop. The main design principles for heat exchangers made
from cellular metals are [2]: high conductivity of metal (preferably Cu, Al), turbulent flow of fluid to enhance local heat transfer; and minimized low-pressure drop
between fluid inlet and outlet. For any system there is a trade-off between heat flux
and pressure drop. Optimum can be achieved with cell sizes in the mesorange and
with relative densities of order r 0.2. [16]
5 Material Properties
(6)
where h is heat transfer coefficient, L is the thickness, and l is the thermal conductivity of the panel. For given Biot number the fire resistance is infinite if the impose fire temperature is smaller than a threshold value TF
TF TC S Bi(TC s T0 )
(7)
where TC and T0 are critical and initial temperatures of the panel, respectively. The
Biot number for a panel made of metallic foam is always higher than that one for a
corresponding metal sheet of the same weight. For example, it takes 8 min to heat
up the opposite side of a Cymat sheet (density 160 kg m 3, geometry 24 cm q
30 cm q 5.5 cm) from room temperature up to 500 hC using oxy/acetylene torch
(theoretical temperature 3100 hC) rotating at 15 cm distance from the foam front
side. The burn-through time for corresponding aluminium sheet of the same
weight (density 2700 kg m 3, geometry 24 cm q 30 cm q 0.34 cm) is only 23 s
[11]. This illustrates the opportunities of metallic foams for this application.
5.3.3
Acoustic Properties
225
In general, the three basic cases can be distinguished with respect to noise attenuation materials:
x
materials for sound insulation: they keep airborne sound out of the protected
space,
materials for sound absorption: they reduce or fully absorb the sound intensity
within the room where the sound is generated,
materials for damping: they reduce the amplitude of the structure acoustic response and hence the sound intensity born by the structure.
In the following the physical principles that determine the potential use of cellular
metals for sound management purposes will be briefly reviewed considering all
three mentioned cases. Guidelines for material selection and procedures ensuring
successful practical applications will be suggested.
Mass controlled
Resonance controlled
Stiffness controlled
5.3.3.1
226
aw
sl
no
s
ma
io
ns
te
Ex
Coincidence dip
Effect of damping
Critical
Frequency Hz
frequency
Principal dependence of sound reduction index of a panel as a function of
frequency for given material and structure geometry [17].
Figure 5.3-6.
5 Material Properties
At very low frequencies the sound reduction is controlled by the stiffness of the
panel. In this case the application of cellular metals can be very attractive because
of their excellent stiffness to weight ratio (see parameter E/r3 in Table 5.3-4).
At somewhat higher frequencies the sound transmission is determined by the
natural resonance of the panel depending on the used material, panel size, and
mounting conditions. Also in this frequency range the cellular metals can be useful, because of their better damping properties in comparison with bulk metals
(see also later in this Section). Moreover, they can shift the first resonant frequency
towards higher values and thus extend the stiffness controlled frequency range (see
Table 5.3-4). This can be important when the panels of smaller size are used.
However, for typical sound insulation, the most important is the frequency
range, where the sound reduction is controlled by mass of the panel according to
(9)
where TL0 in dB is the sound reduction index for normal incidence, m is the superficial density in kg m 2 of the soft impervious panel and f is the frequency in Hz.
This is known as Mass Law [17]: the sound transmission loss increases with increasing mass of the panel. Therefore, the application of cellular metals for
sound insulation in this frequency range is highly disadvantageous.
Table 5.3-4. Design parameters for the materials suitable for sound insulation. Thickness h, 1st
resonant frequency f1 of fully clamped plate and critical frequency fc are calculated for the square
plate of given weight (10 kg) and area of 1 m q 1 m. Parameter E/33 demonstrates the bending
stiffness at minimum weight of the panel. AlSi12 aluminum foam of density 500 kg m 3 and
modulus of elasticity 5 GPa was introduced for comparison.
Material
Lead
E
[GPa]
E/r3
[m8.s 2.kg 2]
h
[mm]
11.2
0.01
0.89
204
8.1
0.4
1.2
Reinforced
concrete
23
2.3
1.9
Glass
38
2.5
Aluminium
69
2.7
Aluminium foam
0.5
Plywood
4.2
0.58
Plasterboard
1.5
0.75
Steel
14.3
r
[g.cm 3]
f1
[Hz]
f1*
[Hz]
60000
11
9770
43
4.4
24
4400
95
2.4
4.0
27
3900
108
3.5
3.7
32
3220
130
40
20.0
110
960
434
22
17.2
81
2241
323
13.3
33
3200
131
3.6
1.8
fc
[Hz]
* 1st resonant frequency calculated for the panel size of 0.5 m q 0.5 m q h m in order to demonstrate the size effect
227
228
5 Material Properties
Beside bulk absorbers two types of resonant absorbers are in general use, especially for sound absorption at lower frequencies. The panel absorber is a resonant
system formed by the mass spring combination of the facing panel and the stiffness of enclosed air. The resonant frequency of panel absorber at which the maximum absorption is reached is given by
600
f p
md
(11)
where m is the superficial density of the panel in kg m 2 and d is the depth of air
gap in cm.
The Helmholtz resonator or cavity resonator is essentially a vessel in which the
mass of air in the neck is driven in and out in resonance upon the stiffness of enclosed air volume. The resonant frequency is given by
r
S
(12)
f 55
LV
where S is the cross-sectional area of the neck, L is the length of the neck, and V is
the enclosed volume of air.
The perforated panel often used for protection of porous absorbers can provide
an additional resonance effect, similarly, as it is in a case of Helmholtz resonator.
Such semiresonant performance is found when the percentage of open area of
front panel is below 20 30 %. Above this value, the system behaves primarily as
a simple porous absorber [17]. The faced material itself can be constructed more
sophisticated containing a plenty of small Helmholtz resonators (see Fig. 5.3-7).
The frequency at which the maximum sound absorption is reached can be
tuned by cavity geometry. Such combination usually shifts the maximum absorption performance towards low frequencies, preferably in the range 100 1000 Hz.
Figure 5.3-7. Typical construction of sound absorber: thick glass fiber absorber (1), thin protective and damping foil (2), facing wood panel with resonator cavity (3), and outer design (4)
[18].
229
230
Accordingly, the main parameter for good sound absorption is the permeability
of the absorber. Therefore, cellular metals can be used for this purpose only if they
meet this fundamental requirement. The closed-cell metallic foams are too stiff to
convert sound energy into heat by vibration of their cell walls. According to Eq. (11),
they cannot be effectively applied for panel resonators because of their very low
weight.
Unfortunately, the experimental measurements performed in impedance tube
on various types of cellular metals [19,20] are in a good coincidence with these discouraging expectations (see Fig. 5.3-8). The best sound absorbing performance was
obtained using Access-sponge metals having open-cell structure and thus very low
flow resistance [21]. Such structure absorbs the sound energy mainly by viscous
losses as the sound pressure wave pumps air in and out of the interconnected
pores. Alporas closed cell foam (made by melt-route process) [22] is manufactured
in large blocks from which the absorbers are machined. Machining results in
rough open-pore surface that slightly improves the sound absorption. Alulight
closed-cell aluminum foam [9] is prepared in net shape and therefore is covered
with a surface skin. This relatively flat and pore free skin reduces the sound absorption performance almost to zero: the sound is entirely reflected.
Figure 5.3-8. Sound absorption coefficient of various types of cellular aluminum foams in
comparison with fiberglass sound absorbing material (directly on rigid background).
Sample
Access sponge
Alporas
Alulight
Fiberglass
Thickness [mm]
20
10
48
Density [kg m 3]
1060
270
580
40
5 Material Properties
Access material has a good potential to be applied for sound absorption purposes
after optimization of the structure. In a case demonstrated in Fig. 5.3-8 the flow
resistance was too low to provide sufficient viscous losses. However, the manufacturing process allows to adjust the optimum pore size or even to obtain gradient
pore structure in relatively easy and inexpensive way [21]. Figure 5.3-9 demonstrates the effect of the reduction of open area at the surface on the sound absorption performance of this material.
Although the sound absorption performance of cellular metals is not very impressive, it can be significantly improved by optimal opening of close cell structure.
In this case the foams can serve as multiple cavity resonator (see Eq. 12) with
much longer neck than in the case of bulk materials of the same weight. The
variability of the pore size will provide a wider frequency range of good sound absorption than typical resonators. Various methods, such as cutting of surface skin,
sand blasting, compressing, rolling, and drilling of holes were examined to open
the foam structure [19,20]. It was shown that sound absorption of as received Alporas aluminum foam can be significantly improved by partial compression, rolling, or hole drilling. However, the combination of these method offers only a little
additional benefits [19].
Similar results were obtained for Alulight foams covered with surface skin (see
Fig. 5.3-10). The significant improvement of the sound absorption was reached already after opening of only 1.4 % of the surface area by drilling of holes. The opening of the surface area can be more economically realized by simple punching of
the foam surface without the need to make the throughgoing holes (Fig. 5.3-11).
Figure 5.3-9. Sound absorption coefficient of Access aluminum sponge without and with
perforated facing Al sheet (air gap of 48 mm behind the absorber, Access sponge: density
1180 kg m 3, thickness 20 mm; facing Al sheet: thickness 1 mm, 61 holes diameter 2 mm).
231
232
Figure 5.3-10. Sound absorption coefficient as a function of opened surface area for Alulight
foam (8.9 mm thick AlSi12 foam, density 580 kg m 3, holes drilled through the sample, 20 mm air
gap between sample and rigid wall).
Surface opening
0.07 %
0.19 %
0.61 %
1.40 %
2.49 %
Number of holes
19
61
61
61
1.5
As expected [23], the increasing thickness of absorber shifts the maximum of the
sound absorption coefficient towards lower frequencies (see Fig. 5.3-12), while the
shape of the curves remains almost unchanged provided, that the density and pore
size do not depend on thickness. At constant sample thickness, the foams with low
density were found to be better sound absorbers than the foams with high densities in the frequency range of 200 1700 Hz [19], however the obtained results are
affected by an increasing pore size with decreasing density. Another experimental
measurement [24] indicates that there exists an optimal pore size for a sound absorption with respect to sample thickness at constant density. This experimental
finding agrees with theoretical calculations made by Wang and Lu [25] for 2D cellular solids of Voronoi structure: the sound absorption performance will be improved with increasing pore size for sample, which is infinitely thick. It implies
that the sound absorption coefficient is a complex function of foam thickness, density, and pore size.
It is not always feasible to increase the thickness of absorber, because of the
weight limitations. The use of a thinner section of absorber spaced by an air gap
gives very similar results to a thick section (see Fig. 5.3-13), however at significantly lower weight. Therefore, the frequency range of maximum sound absorption
5 Material Properties
Figure 5.3-11. Sound absorption coefficient as a function of opened surface area for Alulight
foam (AlSi12 foam 15 mm thick, density 380 kg m 3, diameter of holes 1 mm, holes were punched
into the depth of 8 mm, directly on rigid background).
Figure 5.3-12. Sound absorption coefficient as a function of absorber thickness (Access AlSi
sponge, average pore size 4 5 mm, density 1000 kg m 3, directly on rigid background).
233
234
Figure 5.3-13. Comparison of sound absorpis thicker foam at zero distance from rigid wall
tion for two designs of sound absorbing panels (Access AlSi sponge, density 1000 kg m 3,
average pore size 4 5 mm).
at two constant thicknesses: 1st type is the
foam at given distance from rigid wall, 2nd type
can be tailored by creating an optimum air gap between the absorber and rigid wall
[19,20,24] without need to increase weight. The sound absorption can also be enhanced for wider frequency range by the combination of several foam plates
with an air gap between them [20].
Although the cellular metals cannot provide a high level of sound absorption
itself, they can be attractively used as a facing materials, especially in combination
with fibrous absorbers, provided, that the structure is sufficiently opened. The frequency corresponding to the maximal sound absorption can be tuned by the thickness of foam panel or by changing of pore size. The sound absorbing performance
is then comparable with typical solutions using facing wood panels (see Fig. 5.3-14).
The cellular metals offer additionally very high stiffness at low weight, are selfsupporting and stable at elevated temperature, does not release toxic gases when
subjected to excessive heat, possess high durability under the effects of air flow
or vibration, exhibit sufficient fire resistance and can be recycled. Moreover, as a
relatively new material with impressive surface design, it can be very attractive
for architects and designers of building interiors (see Table 5.3-5).
The potential applications are expected in sterile (antiseptic, dust free) environment, in the structures where nonflammability is important (airplanes, hotels,
commercial, and industrial buildings) or in interiors under difficult conditions (elevated temperature, moisture, dust, flowing gas, and vibrations).
5 Material Properties
Figure 5.3-14. Sound absorption coefficient of fiber glass itself and faced with AlSi12 Alulight
(thickness 13.2 mm, density 330 kg m 3, diameter of holes 2 mm) with Helmholtz resonators
inside (cavity diameter 6.5 mm, length 8 mm). For comparison the sound absorption characteristic of facing sheet of Topperfo acoustic panels TP 16/16/12-2 [18] is presented.
Table 5.3-5.
Sample
Access
Alporas
Alulight
Thickness [mm]
25
10
8.9
Density [kg.m 3]
1060
480
450
Weight [g]
208
38
31
Surface
Cut
Cut
Foamed skin
Opening
As-received
Compressed
Drilled holes
Structure
235
236
Structural Damping
If the structure is subjected to external excitations by sound waves or by mechanical vibrations, the sound can be transmitted or even born by the structure itself.
This is most important in the cases, when the structure oscillates at its resonant
or eigen frequencies. The amplitude of the structure acoustic response and
hence the sound radiation can be dramatically reduced by increasing the damping
of the structure.
Structural damping is owing of internal friction in a material by conversion of
the vibrational energy into heat and subsequent dissipation of heat into the surrounding area. The high level of damping in the structure can be realized either
by utilization of high damping materials or by friction between sliding surfaces
of closely attached structural components interface damping.
The rate of vibrational dissipation or the damping in the structure can be characterized by the loss factor h. The measurement of the loss factor is normally carried out by one of two basic methods [26]:
The frequency response method is used when the damping is high (h j 10 3).
The loss factor can be calculated from the bandwidth Df at the amplitude drop of
3 dB points and resonant frequency fn according to
5.3.3.3
h Df =fn
(13)
The second decay-rate method is used for materials with lower damping. The
loss factor is calculated from the vibration amplitude decay D in dB s 1 at resonant
frequency after instant stopping of exciting force, according to
h 0:0366 D=fn
(14)
The typical structural materials with high strength (for example steel, cast iron,
aluminum alloys) have unfortunately very little damping. On the another hand,
the high damping materials, such as lead, rubber, or soft plastic, have little
strength and cannot be regarded as structural materials [17].
The cellular metals exhibit at least one order of magnitude higher values of the
loss factor h [20]. The dissipation of the vibrations mainly results from the friction
between the attaching surfaces of the cracks appearing in the structure and
partially due to the vibration of the thin pore walls. Thus, the damping can be
enhanced by the reduction of the cell-wall thickness or/and by introducing of
imperfections into the structure. Higher loss factor values are therefore obtained
with for example foams made of casting aluminum alloys, which can be prepared
with very thin cell walls and contain a lot of cracks. If the nonsoluble ceramic
particles, for example SiC, Al2O3, or graphite are additionally introduced into the
structure of metallic cell walls, the damping will be further improved. New sliding
surfaces born in the cell walls in this way are responsible for this behavior (see
5 Material Properties
Figure 5.3-15. Effect of addition of nonsoluble particles on the damping properties of aluminum
foams and steel tubes filled with them (30 mm diameter q 300 mm, foam density 600 kg m 3).
Fig. 5.3-15). Nevertheless, the loss factor of typical cellular metals is still too low
compared with standard damping materials having h in the range 0.01 0.1.
Recently, there is some effort to apply the cellular metals as stiffeners or cores in
metallic hollow parts, castings, or sandwiches. Beside the stiffening and the increased capability to absorb impact energy, also the reduction of noise and vibrations is expected (see Fig. 5.3-16) [27]. In this case the structural damping is significantly affected by the quality of contact between cellular metal and bulk metallic
shell. If the attached surfaces can slide on each other under pressure an additional
interface damping will be obtained. The damping effect is produced by dry friction,
which can provide very effective attenuation of excessive vibrations.
Figure 5.3-16. Frequency spectra of reciprocal mass for a hollow profile and a profile foamed
with various types of Alulight foams (steel tube 22 mm diameter q 245 mm filled with 18 mm
diameter q 245 mm foams of density 700 kg m 3).
237
238
Figure 5.3-17.
Figure 5.3-18. Effect of the foaming on the values of resonant frequency and damping characteristics of hollow steel profile.
5 Material Properties
Figure 5.3-19. Dependence of the first resonant frequency on the apparent density of Alulight
foams: longitudinal vibrations of AlSi12 foam 25 mm diameter q 300 mm (left axis) transverse
vibrations of Al 99.7 foam 140 mm q 140 mm q 8.5 mm (right axis).
Figure 5.3-20. Property gain obtained by use of material was used (weight 390 kg, self deflecaluminium foam for stiffening of hollow mation 4 mm, first resonant frequency 240 Hz,
chine table (800 mm q 720 mm q 250 mm) damping at first resonant frequency 0.15 %)
[28]. As a reference the table made of bulk
239
240
Beside mentioned approaches there is also a possibility to avoid the effect of resonant vibrations by stiffening of the structure. The stiffening arrangements do not
damp the resonances, they merely shift them towards higher frequencies. If the
resonances can be shifted to frequencies that will not be excited during normal operation of the equipment this solution to the problem of reducing vibrations may
be very efficient.
The resonant frequencies generally depend on the modulus of elasticity E and
the density r of the material of which the component is made and, of course, on
its geometry. For example, the first resonant frequency f1 of longitudinal vibration
of cantilever is
s
3:52
EI
(15)
f1
2p
rSL4
where I is area moment of inertia of beam cross section S and L is the length of the
beam.
To shift a frequency either the geometry or material ought to be altered. The
changes of the part's geometry are often not possible. The altering of the material may not be always successful at least in the case of metallic parts, because
E/r ratio of many metals is almost the same. Here the cellular structure can be
very attractive, while both the modulus of elasticity as well as the density can be
varied without changing the outer dimensions of the component (see Fig. 5.3-19).
This feature can be very effectively utilized for shifting the resonant frequencies in
hollow machine components by filling them with cellular metals. Figure 5.3-20
shows the effect of the utilization of aluminum foam as a filler in hollow machine
table. Beside stiffening and improving of the damping a significant shift of the first
resonant frequency towards higher values has been achieved [28].
References
5 Material Properties
12. Norsk Hydro, Data Sheets, Hydro Aluminium Ltd., Sunndalsora, Norway, 1995.
13. L. J. Gibson, M. F. Ashby, Cellular Solids,
Pergamon Press, Oxford, 1988.
14. E. S. Park, S. D. Poste, US Patent 4 808 558,
1989.
15. T. J. Lu, C. Chen, in Metal Foams and Porous
Metal Structures, J. Banhart, M. F. Ashby,
N. A. Fleck (eds), MIT Verlag, Bremen 1999,
p. 391.
16. A. G. Evans, J. W. Hutchinson, in Metal
Foams and Porous Metal Structures, J. Banhart,
M. F. Ashby, N. A. Fleck (eds), MIT Verlag,
Bremen 1999, p. 45.
17. P. D. Wheeler, in Industrial and Machinery
Noise Control Practice, Institute of Sound &
Vibration Research, University of Southampton, UK 1978, p. 11.1.
18. n'H Akustik Design AG, Data Sheet,
Lungen, Switzerland, 1999.
19. T. J. Lu, A. Hess, M. F. Ashby, J. Appl. Phys.
1999, 85, 7528.
20. J. Kovacik, P. Tobolka, F. Simancik, in
Metal Foams and Porous Metal Structures,
J. Banhart, M. F. Ashby, N. A. Fleck (eds),
MIT Verlag, Bremen 1999, p. 405.
241
6
Modeling and Simulation
F. G. Rammerstorfer, T. Daxner, and H. J. Bohm
The modeling and simulation of cellular metals is a field of research that has attracted considerable interest from researchers, research groups, foam manufacturers, and industrial end-users. Two major directions of research can be distinguished: the design of metallic foams on the one hand and the design of components made of cellular metals on the other.
Modeling and Simulation for Material Design
From the point of view of the manufacturers, the design of mechanically efficient
metallic foams is the paramount aspect of foam modeling and simulation, the aim
being the development of optimum cellular metals. The most important design
variables are the production technologies and the choice of the metal making up
the solid phase of the foam. Besides the apparent density and the properties of
the metal, the mechanical behavior of metallic foams is determined by the geometrical arrangement of voids and solid regions in the material, which will be referred
to here as the microgeometry. Consequently, a considerable number of studies have
aimed at exploring the connections between the microgeometry and the mechanical properties of cellular materials and thus belong to a research field known as
micromechanics of materials.
At the length scale of the voids, the deformation of cellular materials tends to be
dominated by local mechanisms that must be accounted for in any modeling effort.
Because detailed descriptions of the microgeometries of large structures or components made of foam are far in excess of present capabilities, the most fruitful approaches for simulating the thermomechanical behavior of metallic foams have
aimed at studying representative regions of appropriate model materials in detail.
Most commonly, actual foams are approximated by periodic microgeometries that
can be described via unit cells subjected to appropriate boundary conditions. Alternatively, a geometrically detailed microregion may be embedded in a larger region
for which a much simpler description is employed, leading to embedded cell models. A discussion of such micromechanical approaches is presented in Section 6.1.2.
Microgeometrical models covering a wide range of complexity can be studied
with the above methods to give predictions for the overall mechanical behavior
243
244
6.1
In the course of the last decade the modeling and simulation of cellular materials
have gained in importance. Well-established representatives of this group, such as
honeycombs or polymer foams, as well as recently developed materials, such as
metallic foams, have been the subjects of various studies. Cellular materials,
which consist of a solid skeleton in the form of struts and/or cell walls and a
high volume fraction of voids, are highly inhomogeneous. This heterogeneity
leads to thermomechanical responses that are markedly different from those of
bulk solids and it gives rise to material properties that have made cellular materials
attractive for many engineering applications. It also presents an obvious target for
modeling studies aimed at gaining an improved understanding of the mechanical
behavior of cellular materials and structures made of them.
245
246
6.1.1
Motivation
Metallic foams, which are the subjects of the present discussion, have entered the
stage of practical application, their most important mechanical properties being excellent energy absorption capacity as well as high specific strength and stiffness.
Studying their mechanical behavior has attracted considerable research interest,
both from the theoretical and the applications points of view.
Simulations of the mechanical responses of cellular metals may be carried out at
different length scales. In micromechanical approaches the inhomogeneous structure of these materials is accounted for at the level of individual cells, cell walls,
struts, and vertices either via statistical arguments or by discrete geometrical models. The corresponding length scale, which is of the order of the size of individual
cells, is called the microscale in the following. Studies at this level provide information on the local deformation and load-transfer behavior that can then be correlated
to the mechanical behavior of the material at the structural level. A good understanding of the mechanics of cellular metals on the microscale is especially important for identifying advantageous microgeometrical parameters for materials design and development. Microscopic models for metallic foams are the focus of
Section 6.1.2.
For studying samples and components that are, say, two or three orders of magnitude larger than the size of an individual cell, only the overall thermomechanical
behavior is of interest and local details of the metallic foam do not have to be accounted for explicitly. In other words, at the macroscale the foam is treated as a
homogeneous continuum rather than an inhomogeneous medium. Modeling approaches appropriate for describing metallic foams at the macroscopic level,
which are often rather application specific, will be discussed in Section 6.1.5. In
addition, questions involving the spatial variations or gradients of cell sizes and
shapes within a given sample or structure may be studied at length scales that
are intermediate between microscale and macroscale. Some simple models involving such a mesoscale will be presented in Section 6.1.4, while Section 6.2 is
largely devoted to mesoscopic approaches.
In the following some important aspects of the present state of the modeling and
simulation of metallic cellular materials are discussed. It should be noted, however,
that no attempt is made to give a full overview of this research topic. For comprehensive information on the behavior of cellular materials and of its modeling the
reader is referred to the books by Gibson and Ashby [1,2] and to an overview by
Weaire and Fortes [3]. A recent review of the mechanical behavior of metallic
foams was given by Gibson [4].
6.1.2
From the point of view of micromechanical modeling, cellular metals fall into three
groups: honeycombs, which can be studied by 2D models, open-cell foams, the
solid scaffold of which is dominated by beam-like members, and closed-cell
foams, in which membrane- or shell-like cell walls are present. All of these materials typically show a limited elastic range and their mechanical behavior tends to be
dominated by local deformation mechanisms such as bending, buckling, plastic
yielding, and fracture of cell walls and struts. These deformation mechanisms,
in turn, are highly sensitive to details of the microgeometry, which, in practice,
may be quite regular for some honeycombs, but tends to be complex and highly
non-uniform in the case of metallic foams.
The mean field and HashinShtrikman-type methods that play an important role
in continuum micromechanics of composites and materials with small volume
fractions of pores [5], have seen only limited use for cellular materials. Hashin
Shtrikman upper bounds for the elastic moduli can be evaluated for macroscopically isotropic cellular materials by prescribing vanishing stiffness for the void
phase (the lower bounds, however, vanish trivially). Typically, the elastic moduli
of metallic foams lie considerably below these upper bounds, but for honeycombs
it was possible to identify microgeometries that realize the upper bounds [6].
As a consequence, most micromechanical studies of cellular materials in general
and of metallic foams in particular have been based on discrete microgeometrical
models. Owing to the high geometrical complexity and irregularity of actual metallic foams such model microgeometries tend to be highly idealized. One common
modeling strategy is based on studying periodic model foams, the thermomechanical behavior of which is fully described by appropriate unit cells. Alternatively, cells
or geometrical units may be studied in isolation without requiring them to be
space filling. A third type of approach employs a geometrically fully resolved microregion (core) that is embedded in a much larger region in which the microgeometry is not resolved and smeared-out material behavior is used (embedded cell
models). Discrete microgeometrical modeling approaches facilitate the controlled
variation of selected geometrical parameters in order to assess their effect on the
mechanical response. These modeling strategies, which have well-established
equivalents in continuum micromechanics of composite materials, allow studying
both the local deformation mechanisms and the corresponding overall behavior,
which can be obtained by homogenization.
247
248
Classification of Microgeometries
This section represents an attempt at systematically classifying the most important
types of cellular microgeometries that have been or may be used for unit cells or in
the core regions of embedded cell models. The main classification criteria are the
dimensionality (2D versus 3D) and the microscale morphology (regular periodic,
perturbed periodic, random, and real structure arrangements) of the geometrical
models.
2D discrete microgeometry models are directly applicable to investigating the
mechanical behavior of honeycombs. Owing to their relative simplicity they have
also been used as tools for studying metallic foams in a qualitative way (in the following planar models are generally referred to as honeycombs). The baseline
regular morphology are periodic hexagonal honeycombs, which show in-plane elastic isotropy and, in the context of 2D liquid foams, may be viewed as surfaces of
minimal surface energy. Because hexagonal honeycombs (albeit of reduced symmetry) are applied routinely as sandwich cores, cellular metals of this type have
been closely studied both experimentally and analytically [21]. Alternative regular
microgeometries for 2D cellular materials were also discussed in the literature
[11,12].
In contrast to the regular microgeometry of idealized models, real cellular metals
are subject to morphological defects that tend to lead to softer and weaker mechanical responses. Periodic 2D models have been an important means for studying the
influence of imperfections such as curved or corrugated cell walls [16,17,22], perturbed cell shapes [23], and non-uniform material distribution between cell walls
and cell vertices (Plateau borders) [24] as sketched in Fig. 6.1-1. Within such honeycomb models local defects can be introduced into cell walls [17,25,26], individual
cell walls can be eliminated [26,27], vertices and the adjoining cell walls can be removed to leave large voids in the honeycomb structure [26,28], and cells or groups
of cells can be filled with solid material to assess the influence of nodal inclusions
[26], Fig. 6.1-2. In addition, periodic arrangements of large and small cells [29] may
be used to study influences of relative cell sizes.
Random honeycomb arrangements can be generated on the basis of Voronoi tesselations [17,25,27,30], which may be interpreted in terms of geometries that
emerge when the fronts of bubbles growing with the same linear rate from ran6.1.2.2
Schematic representation of a
honeycomb containing imperfections in the
form of a large hole, cells filled with material, fractured cell walls, and geometrical
perturbations (from left to right).
Figure 6.1-2.
249
250
251
252
The above concept can also be used to follow the evolution of the yield surface as
a given macroscopic loading path is traced. To do this, the loads are applied incrementally and complete unloading is carried out after each step followed by the evaluation of a new yield surface. This may again be done by superimposing elastic
solutions. It should be noted, however, that the application of linear superposition
analyses for such purposes is limited to the small deformation regime. In cellular
metals geometrical nonlinearities (caused, for instance, by marked bending of cell
walls) may give rise to noticeably nonlinear overall responses even before yielding
sets in.
As a foam is subjected to increasing compressive stresses a load will be reached
at which the first collapse of a cell occurs. In metallic foams the initial collapse of a
cell under these conditions tends to be followed by the growth of the collapsed region, which typically takes place at stress levels that show only limited variation,
giving rise to a so-called plateau region in the overall stress versus strain behavior.
When a considerable percentage of cells have been consumed the foam densifies,
the plateau region ends, and a much stiffer response sets in. The accumulation of
considerable compressive strains at nearly constant stresses is of particular interest
for cellular metals that are to be employed for impact energy absorption. Collapse
stresses under uniaxial loading have been studied for many cellular morphologies
and for various imperfections. For general load cases the initial collapse stress
states form an envelope surrounding the initial yield surface, which may be called
a collapse surface. For the point-wise evaluation of collapse surfaces fully nonlinear
analyses along individual load paths are required.
Relatively little work has been reported on the modeling of fracture of metallic
foams, which is the dominant failure mechanism under macroscopic tensile loading. Embedded cell models provide a flexible approach for studying cracks in idealized cellular metals, see Section 6.1.3.8.
6.1.3
Simone and Gibson [24] developed unit cells for hexagonal honeycombs and tetrakaidecahedral foams in which Plateau borders were modeled by appropriate
curved and parallel regions. Their results indicate that the distribution of material
in the cell walls has little influence on the Young's modulus and only a moderate
effect on the uniaxial yield strength of closed-cell metallic foams. A different approach was followed by Chen et al. [17], who studied honeycombs with wall thicknesses that increase linearly from the middle of the walls to the vertices. This simplifies the mathematics of the problem sufficiently for analytical solutions for the
overall yield surfaces of the models to be obtained.
Investigations of the deformation patterns of honeycombs in the large strain regime, which determine the materials' energy absorption capacity, can also be carried out with unit-cell models. In such simulations the initiation of densification
can be identified from the occurrence of self-contact of the surfaces of the collapsed
voids, the modeling of which requires special provisions in terms of the unit-cell
geometries. Unit cells with all boundaries running within the cell walls have
been found to answer well, see Fig. 6.1-5, which shows symmetric and periodic deformation modes before the onset of densification. The main advantage of unit
cells of the above type is that all free surfaces of the voids face inwards, which allows a straightforward use of contact algorithms. Obviously, such models can also
be adapted for studying the dependence of the deformation patterns on prescribed
distributions of mass between cell walls and vertices.
Drainage of the material into Plateau borders reduces the cell wall thickness,
which is the governing parameter for the strength and the stiffness of the arrange-
Figure 6.1-5.
253
254
ment. This is especially important in the case of foams of very low density, in
which the cell wall thicknesses approach zero. At higher densities, however, the redistribution of material towards the cell vertices and, therefore, into regions subjected to higher bending moments can be beneficial for the overall properties of
foams [17,24].
As more material is accumulated in the vertices, the plastic hinges under bending tend to form closer to the mid-region of the cell edges, so that the size of the
nearly rigid regions around the vertex is increased. Although such a zone can rotate during compression it still takes up more space if the distance between the
hinges and the vertex centers is increased. Together with changes in the folding
kinematics this causes the cell walls to contact each other at lower strains than
is the case for microgeometries with less pronounced plateau borders. This effect
dominates the energy absorption potential as shown in Fig. 6.1-6, Fig. 6.1-7, and
Fig. 6.1-8. Even in cases where the collapse and pre-contact loads are increased
by the material redistribution, the decrease of the usable deformation length reduces the amount of energy that can be dissipated by plastic deformation before
the densification regime is reached.
Microscale fluctuations of the material distribution in cellular materials may, on
the one hand, be due to thickness variations within individual cell walls as discussed above in connection with Plateau borders or, on the other hand, be caused
by the individual cell walls in a microgeometry having different thicknesses. The
latter problem was addressed by Grenestedt and Bassinet [36], who developed a
3D tetrakaidecahedral model of a closed-cell foam that contains a total of 112
Figure 6.1-6. Predicted nominal strain at first cell wall contact as a function of the material
distribution between walls and cell vertices (Plateau borders) for three different apparent
densities.
Figure 6.1-7.
Figure 6.1-8. Predicted normalized absorbed energy (up to first cell wall contact) as a function of
the material distribution between cell walls and cell vertices (Plateau borders) for three different
apparent densities.
cell walls, the thicknesses of which could be assigned individually. The results
show that the stiffness of regular closed-cell arrangements is rather insensitive
to the presence of cell walls of different thickness. This was explained by noting
that such microgeometries deform primarily by cell wall stretching, which is
less sensitive to wall thickness effects than bending modes induced, for example,
by corrugated cell walls.
255
256
foams, however, curved and corrugated (wavy) cell walls typically are present, see
Fig. 6.1-4.
Grenestedt [16] studied the influence of wavy imperfections of the walls on the
elastic stiffness of cellular solids and found that the bulk modulus decreases by
some 50 % when corrugations with an amplitude of twice the thickness of the
cell walls are present. Simone and Gibson [22] documented the detrimental effect
of curved and corrugated cell walls on the stiffness and strength of regular honeycombs and perfectly tetrakaidecahedral unit-cell models. The Young's modulus was
shown to be more adversely affected by these imperfections than the collapse
stress. Chen et al. [17] gave analytical expressions for the yield surface of regular
honeycombs with wavy cell walls. They found that corrugations significantly reduce the hydrostatic yield strength of honeycombs, whereas the deviatoric yield
stress is hardly affected. Evidently, cell-wall bending is activated in addition to
membrane deformations under overall hydrostatic loading.
a)
b)
c)
d)
Figure 6.1-9. Predicted yield surfaces for regular and irregular honeycombs with different types and
degrees of imperfections [29]. Dashed and dash-dotted lines denote elastic buckling stress states.
The influence of corrugated and curved cell walls on the onset of nonlinear behavior was studied by Daxner et al. [29] by evaluating initial yield surfaces.
Figure 6.1-9 shows such curves that represent overall stress states that give rise
to the local onset of yielding or to local elastic buckling in regular and perturbed
honeycombs that have curved and wavy cell walls of varying curvature. In perfect
regular honeycombs in-plane hydrostatic compression leads to pure membrane
stresses in the cell walls, which give rise to an elongated overall initial yield surface
in the s x s y stress plane. For low-density, perfect honeycombs, this yield surface is
truncated by the failure surface for elastic cell wall buckling, see Fig. 6.1-9a, which
shows a kink owing to a change of the buckling mode. This type of behavior is
maintained in the presence of small corrugations, but when the amplitudes of wiggles exceed about 5 % of the wall thickness yielding alone determines the onset of
nonlinearity.
The influence of wiggles and curved cell walls on the yield surfaces of honeycomb microgeometries that are irregular from the outset is much less pronounced,
see Fig. 6-1.9bd. Perturbing the geometry by randomly displacing cell vertices
causes the yield surface to shrink even more than the presence of severe wiggles,
see Section 6.1.3.3. Corrugated or curved cell walls generally have little effect on
the uniaxial yield strength, but severe wiggles can reduce the in-plane hydrostatic
yield stress by some 35 %. Curved cell walls typically give rise to similar effects as
wiggles, see Fig. 6.1-9c, but cell walls incorporating only slight curvature can lead
to minor increases of the in-plane hydrostatic yield stress.
257
258
Honeycomb geometries for studying the interaction of cells of different sizes [29]:
Regular hexagonal arrangement (HC, left), a cluster of small cells surrounded by big cells (SBS,
center), and big cells surrounded by small cells (BSB, right).
Figure 6.1-10.
curves. If, in contrast, cell wall buckling and extensive layer-wise collapse perpendicular to the loading direction are the dominant mechanisms, the stress strain
relationship shows marked oscillations on account of the sequential collapse of
cell rows that are incapable of accommodating subsequent deformation by sliding. Figure 6.1-11 shows deformation modes of the above types for the case of an
imperfect hexagonal honeycomb, with the corresponding stress versus strain
curves being given in Fig. 6.1-12.
In addition to these general observations on collapse modes some specific rules
for the interaction of small and big cells can be identified. Comparing the deformation patterns developed by arrangements SBS and BSB, see Fig. 6.1-13, one can
observe that in the latter the clusters of small cells do not contribute very much
to the overall deformation and may become obstacles in the densification regime.
The framework of small cells surrounding the large ones in model SBS, however,
provides more uniform stiffness and deforms more evenly. In general, clusters of
small cells tend to be detrimental to the deformation and energy absorption potential of cellular metals as they typically deform less than their bigger neighbors, the
longer struts of which confer lower bending stiffness and strength.
Figure 6.1-11.
259
260
Predicted overall stress strain relationships of an imperfect hexagonal honeycomb [29] corresponding to the deformation modes shown in Fig. 6.1-11.
Figure 6.1-12.
Figure 6.1-13.
These observations lead to the conclusion that whenever smooth stress versus
strain relationships are desired, the cell size distribution should be as uniform
as possible and the cells should be small compared to the sample or structure to
either prevent localization or to limit its detrimental effects.
Experiments by Prakash et al. [28] have shown that the filling of some cells leads
to local strengthening of honeycombs, increases the elastic modulus and the degree of strain hardening, but reduces the densification strain. The opposite effect
is caused by removing individual cells or whole cell clusters as demonstrated by
Guo and Gibson [45] in a FE study on intact and damaged honeycombs. They reported on correlations between the undamaged cross-sectional area perpendicular
to the loading direction and the elastic buckling load as well as the plastic collapse
strength. The interaction between separate defects of this type was found to have a
range of about ten cell diameters.
Chen et al. [26] investigated the influence of solid inclusions as well as holes on
the stiffness and on the uniaxial and in-plane hydrostatic yield strengths of perfect
honeycombs and of 2D arrangements in which some 5 % of the cell walls were randomly fractured. Solid inclusions were shown to lead to minor increases of the
elastic stiffness and to have a negligible effect on the uniaxial and in-plane hydrostatic yield strength of both the otherwise perfect and the fractured honeycombs.
Because the solid inclusions introduce additional mass, however, the specific properties were negatively affected. Large holes were found to induce cell wall bending
in otherwise perfect honeycombs, leading to significant reductions in the bulk
modulus and the hydrostatic yield strength. For prefractured honeycombs the decrease in overall stiffness due to holes could be estimated from the reduced overall
relative density of the honeycomb.
261
262
Unit cell for a periodic real structure honeycomb model [32] adapted from a
micrograph [16]. Note the wide range of cell sizes, cell shapes, and cell wall imperfections.
Figure 6.1-14.
Figure 6.1-15. Overall yield and collapse surfaces [32] predicted for the real structure honeycomb
model presented in Fig. 6.1-14. The collapse surface is not evaluated for load cases for which a
distinct maximum could not be found in the overall stress versus strain relationship.
263
264
isotropic hardening [47]. It assumes that the origin and the shape of the yield surface stay the same, but its size increases driven by the accumulated volumetric
plastic strains. In the Crushable Foam model [48] mixed hardening is assumed,
where the yield surface keeps the same shape with increasing size (isotropic hardening), but shifts its origin (kinematic hardening) so that the hydrostatic tensile
yield stress remains constant.
Micromechanical analyses based on the real structure geometry shown in
Fig. 6.1-14 paint a somewhat more complex picture. Figure 6.1-16 displays predictions for the evolution of the overall yield surface corresponding to progressive uniaxial compressive loading in the x-direction. At several points during the loading
sequence the analysis was interrupted, the model was unloaded and then subjected
to linear superposition analyses to estimate the onset of nonlinear behavior. This
way the yield surface corresponding to the actual state of hardening can be generated. The initial hardening behavior was predicted to be essentially kinematic, the
yield surface being shifted in the direction of the applied stress without major
changes in shape or size. At elevated compressive stresses, however, the evolving
yield surface changes its shape and contracts as well. Because Fig. 6.1-16 is
based on the superposition of solutions from linear analyses that cannot account
for the geometrical nonlinearities, which may be expected to play an increasing
role as compressive collapse is approached, its reliability in the latter regime is
not fully clear. Accordingly, the results can be interpreted as clearly favoring kine-
Figure 6.1-16.
matic hardening models for low inelastic strains and hinting at possible changes in
the hardening behavior in the high strain regime.
As mentioned before, planar models are directly applicable to honeycombs, but
are not necessarily directly transferable to real foams, which require 3D simulations for quantitative predictions. Denzer developed 3D closed-cell microgeometries [32] based on a body-centered cubic arrangement of tetrakaidecahedral cells,
using symmetry boundary conditions to limit the computational costs. Initial overall yield surfaces under compressive loading were evaluated for configurations with
perfect cell geometry as well as for models with statistically perturbed vertex positions, see Fig. 6.1-3 and Fig. 6.1-17. The general shapes of these yield surfaces in
principal stress space are ellipsoid-like and their major axes are aligned with the
hydrostatic axis, see Fig. 6.1-18. This can be explained by the stress states induced
in the cell walls, which are mainly membrane-like for hydrostatic loads (even
though regular tetrakaidecahedra are not elastically isotropic) whereas uniaxial
loads tend to give rise to bending close to the vertices. These effects are less pronounced for the imperfect models, where the aspect ratio of the yield surface
tends to be lower. This behavior is qualitatively similar to predictions obtained
from honeycomb models, but the yield surfaces for the perfect tetrakaidecahedra
are less elongated along the hydrostatic axis, see Fig. 6.1-9.
By projecting selected points of the yield surface shown in Fig. 6.1-18 (left)
onto the equivalent stress versus mean stress plane a standard representation is
obtained, see Fig. 6.1-19, that can be readily compared with other results. The projected points can be seen to be scattered around an ellipse the major axis of which
coincides with the mean stress axis, the hydrostatic yield stress being about 2.5
Figure 6.1-17. Irregular tetrakaidecahedral model for a closed-cell foam [32]. As can be seen from
the planar section, perturbations were introduced only in the core of the model to allow for the
use of symmetry boundary conditions.
265
266
Yield surfaces predicted for regular (Fig. 6.1-3) and irregular (Fig. 6.1-17)
tetrakaidecahedral closed-cell foams [32].
Figure 6.1-18.
Projection of the yield surface shown in Fig. 6.1-18 (left) onto the von Mises
equivalent stress versus mean stress surface.
Figure 6.1-19.
times larger than the uniaxial yield stress. It may be noted that many material laws
for foams use yield surfaces that give rise to ellipses when projected to the equivalent stress versus mean stress plane, among them the model of Deshpande and
Fleck [47].
Figure 6.1-20.
267
268
Figure 6.1-21. Macroscopic force versus displacement curves of CT specimens made of open-cell
aluminum foams predicted by embedded cell models, see Fig. 6.1-20, for different effective
densities and different aluminum alloys [35].
Figure 6.1-22. Macroscopic crack resistance curves of CT specimens made of open-cell aluminum foams predicted by embedded cell models, see Fig. 6.1-20, for different effective densities
and different aluminum alloys [35].
were obtained. For this purpose, the macroscopic stress intensity factor KI was evaluated at each maximum of the force versus displacement curves according to standard expressions for CT specimens, the crack length increments Da being dictated
by the selected microgeometry. Both Fig. 6.1-21 and Fig. 6.1-22 show results for
three values of the apparent density of the foam and for two aluminum alloys differing in ductility.
Within such a modeling approach parameters such as the mass density of the
foam (which determines the void volume fraction), the size of the cells (which influences the fracture behavior in terms of an internal length scale), the yield
strength, and the strain to fracture of the bulk material (which govern the ductile
damage and failure of the struts) can be varied easily in order to study their respective influences on the fracture behavior of open-cell metallic foams.
6.1.4
Foams are not only heterogeneous in terms of consisting of solid regions and
voids, but in many cases also show an inhomogeneous distribution of their apparent density, mean cell size and other geometrical parameters. These variations correspond to length scales larger than that of the voids (the microscale) and smaller
than that of the sample or component (the macroscale) and are, accordingly,
termed mesoscopic inhomogeneities in the following. Fig. 6.1-23 (left) shows
such density variations in a cross section of a foam sample.
Shim et al. [51] introduced uniaxial lumped massspring mesoscopic models for
the simulation of the dynamic uniaxial crushing of foam samples with uniform apparent density. Systems of elastic-plastic springs were employed by Gradinger and
Rammerstorfer [52] for studying the compressive force displacement behavior of
crush specimens with deterministically distributed meso-inhomogeneities. These
two approaches were combined by Daxner et al. [53] to assess the influence of inhomogeneous density distributions on the impact response of foam plates with generic variations in their apparent density, the interaction of density inhomogene-
Figure 6.1-23. Micrographical cross-section of an aluminum foam with three regions of differing
apparent density (left) and sketch of the corresponding mesoscopic simulation model (right) [53].
269
270
ities in directions normal to the orientation of the load being also accounted for,
see Fig. 6.1-23 (right).
The influence of mesoscale inhomogeneities in the apparent density is of particular interest when the metal foam is applied as protective padding. Strong mesoinhomogeneities reduce the energy absorption efficiency, because they lead to strain
localization resulting in a decrease of the initial plateau stress and a less pronounced plateau region [52,53], as can be seen in Fig. 6.1-24. This holds true for
density variations occurring in the loading direction and perpendicularly to it
[53]. Accordingly, for such applications it is desirable to tune foam production technologies for achieving uniform density distributions.
When a metallic foam is treated as a homogenized material, inertia effects [54]
lead to wave propagation phenomena that are controlled by the interaction of foam
density, layer thickness, impact energy and impact velocity [53]. At moderate impact speeds those inertia effects give rise to an impact resistance that only slightly
exceeds the static yield strength, see Fig. 6.1-25. Nevertheless, the dynamic effects
of wave reflection and superposition are noticeable under these conditions, as is
evident in Fig. 6.1-26, where the stacking order of foam layers with different density is shown to significantly influence the time history of impact stresses. The results obtained with simulations that treat cellular metals as homogenized materials, however, will have to be compared to dynamic tests to assure that no artificial
inertia effects are introduced by the homogenization approach. Possibly, the mass
density used for the homogenized material may have to be adjusted to closely fit
the experimental results for dynamic events.
Figure 6.1-24. Stress strain diagrams prea meso-homogeneous plate, the open circles
dicted for foam plates that consist of layers of denote a two-layer structure and all other cases
different density while maintaining the same
are three-layered arrangements [53].
overall density. The solid line corresponds to
Figure 6.1-25. Predicted influence of impact velocity and foam density on the initial stress
response s dyn normalized by the static collapse stress s 0,stat [53].
Figure 6.1-26. Predicted contact stress between an impacting mass and a foam cube composed
of two equally thick layers of different densities [53]. The two possible stacking orders of the layers
were simulated.
271
272
6.1.5
From the perspective of modeling samples and structural components made of cellular metals, discretization down to the level of individual cells is not only infeasible with current computer and software technology, but also appears to be impracticable with regard to the required modeling effort. As a consequence, macroscopic
constitutive descriptions of metallic foams have been developed. The main condition for their use is that the overall dimensions of the sample or structure must be
significantly larger than the dimensions of single foam cells.
Structural analysis of cellular metals in the elastic range is relatively straightforward in that only the homogenized elasticity tensor must be known. For isotropic
metallic foams, accordingly, two elastic moduli have to be calibrated; more material
parameters are required for foams of lower symmetry. Experiments, however, have
shown that metallic foams may not display a distinct elastic regime under compressive loading, but rather begin to deform irreversibly at stresses that are low
compared to their collapse or plateau stress. Thus, it is not fully clear to what extent
cellular metals can be treated as linear elastic materials in structural analyses.
Practically relevant constitutive models for metallic foams, accordingly, must be
capable of accounting for nonlinear macroscopic material responses. If, for example, the FE method is to be used in the engineering analysis of components made
of cellular metals, constitutive descriptions are required at the integration point
level, which may have to be coded specifically [55].
Macroscopic yield surfaces appropriate for metallic foams typically can be represented by ellipses in the mean-effective stress plane. The yield surface proposed by
Deshpande and Fleck [47] is symmetrical with respect to compressive and tensile
stress states. Furthermore, it evolves by uniform scaling in all directions (isotropic
hardening) and uses an associated flow rule, which implies that the plastic Poisson
ratio is generally not equal to zero. In contrast, the Crushable Foam material model
[48] is characterized by a constant tensile hydrostatic yield stress, which causes the
origin of the yield surface to move along the hydrostatic axis as the yield surface
expands during hardening. The non-associated flow rule governing this material
model is formulated to always prevent a plastic Poisson effect. The accumulated
plastic volumetric strain is the internal variable driving the hardening process.
For additional macroscopic constitutive models proposed for metallic foams see
[56 58].
For the above material models additional parameters determining the shape of
the yield surface and the hardening behavior must be provided by the user, requiring information on the mechanical behavior of the material under multiaxial loading conditions. Since multiaxial experimental data [47,59] is scarce, the user very
often must rely on appropriate assumptions, which can be derived either from
microstructural FE simulations or via parameter identification techniques (that
is, by minimizing the discrepancy between simulations and experimental results).
Fortunately, applications that are dominated by hydrostatic loading conditions
are rare. To illustrate this point, in the next section a case study will be pre-
sented in which the uniaxial compressive stress strain relationship is the most
important material characteristic: the impact of a large sphere on an aluminum
foam pad.
Figure 6.1-27.
273
274
Figure 6.1-28. Schematic view of an impact event (left) and predicted deformed configuration
after the impact (right).
ing due to fracture of cell walls may also occur when stress states enter tensile regimes.
The evolution of local stress states can be visualized in the mean versus equivalent stress diagram used commonly for describing pressure dependent material
laws. Figure 6.1-29 shows such stress paths for three different material points.
These stress paths lie mostly between the axis of uniaxial compression (inclination
3:1) and the vertical axis that represents pure shear, indicating that these two types
of loading will dominate the local deformation of the foam. The `Axis, top' line in
Fig. 6.1-29 is associated with a point under the center of the impact mass. It is surrounded by an ellipse, which is the projection of the Crushable Foam yield surface
corresponding to the highest value of the hardening parameter, namely the accumulated volumetric plastic strain. Uniaxial stress states can be seen to prevail in
the impact event considered here, indicating that the simulation should not be
very sensitive to changes in the hydrostatic parameters of the material. This is comforting, since only limited information is available on the behavior of foams under
hydrostatic pressure. Furthermore, the principal stress axes maintain roughly the
same orientation throughout the impact event, so that the selected hardening
model does not play a major role. These findings narrow down the requirements
on improved material laws for crash simulations involving a comparatively thin
padding and a large impactor.
111111111111
000000000000
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
000000000000
111111111111
00
11
000000000000
111111111111
00
11
000000000000
111111111111
00
11
2.5
2.0
1.5
1.0
0.5
0.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
275
80
Load [kN]
276
60
40
20
20
40
60
Displacement [mm]
80
100
Figure 6.1-30.
Folding mechanisms for empty prismatic profile (left) and simplified deformation
kinematics of foam core (right).
Figure 6.1-31.
6.1.6
Metallic foams show some potential for being produced with controlled spatial
variations of their density, introducing inhomogeneities on a mesoscale as defined
in Section 6.1.4. In the case of foams produced via powder metallurgical routes the
foaming process, and, as a consequence, the final density distribution may be controlled by suitable choices of the distribution of the blowing agent and of process
parameters such as temperature and pressure. This suggests employing such
foams as graded materials in space filling lightweight structures designed in analogy to cancellous bone, a natural functionally graded cellular material that displays
increased density in regions of high loading, see Fig. 6.1-32. Reiter et al. [72,73]
presented algorithms for the optimization of composite materials that are based
on procedures for simulating the natural adaptation of bone to applied loads.
Daxner et al. [74] implemented the algorithm of Reiter [73] for a self-adapting
material, providing options for converging towards a uniform distribution of the
local strain energy density or towards the satisfaction of a local yield criterion by
iteratively adapting the local density. The former choice gives rise to configurations
of increased stiffness and the latter to density distributions in which the material
shows uniform safety against plastic failure.
It was shown that strength optimization with material parameters that are typical for metallic foams leads to continuous solutions showing density gradients, see
Fig. 6.1-33, which may indeed be realizable by suitable processing technologies. Attempts to optimize the specific stiffness of foam structures resulted in the formation of discrete structures such as frameworks of struts. This indicates that cellular
metals may not be ideal for stiffness-optimized structures, at least in the presence
of concentrated loads. For sandwich beams with foam cores as studied by Vonach
et al. [75] smooth mesoscopic density distributions were also obtained for stiffness
optimized configurations under distributed loads.
6.1.7
Outlook
277
278
human femur.
Strength optimized symmetric beam with fixed supports under a central load.
Highly stressed regions such as the supports and the lower and upper regions in the middle of
the beam accumulate the highest local foam density [74]
Figure 6.1-33.
In view of the engineering potential of cellular metals and of the research challenges they pose, it appears safe to say that studies of these materials on all length
scales discussed above will remain a fertile field of study for many years to come.
Acknowledgements
References
279
280
6.2
6.2.1
Introduction
Cellular metals are a class of material that offers special properties as described in
Chapter 5. Their successful introduction into industrial application requires the
development of design methods that predict the responses of these materials to
external loads.
The discrete mass distribution of cellular metals causes discontinuities in their
local properties. To determine the response of a highly porous solid to external
loading, such a cellular structure can be modeled as a network of curved beams
(open-cell foam) or shells (closed-cell foam). This model can be analyzed using
available numerical solutions such as finite elements or boundary elements methods. The 3D cellular structure of a real, foamed metal (see Chapter 4) is too complex to be easily handled with such a model due to excessive calculation time.
Another method is to approximate the cellular structure by a continuum model.
Such a homogenization method was used by some of the theoretical studies [1]. In
this approach the cellular structure is homogenized over the whole sample or component, a scale much larger than the typical microstructure.
Theoretical studies [1] suggest that the mechanical properties of an ideally homogeneous foam are scaled by rn. In these relations that are also called scaling laws
the r is the relative density of the foam and n is a constant that depends on the
mechanism governing the deformation of the cell walls. For example, the elastic
modulus of an open-cell foam with low relative density (
r J 0.1) is governed by
bending of struts and scales with r2. At higher relative densities (0.1 J r J 0.3)
the effect of shear and axial deformation of struts shall be considered [2]. Correction of the scaling law shows that the influence of these effects on the final
response of an open-cell foam in the elastic regime is small. Prediction of such
scaling laws and their correction at medium densities are compared in Fig. 6.2-1.
In closed-cell foams, bending of struts is accompanied by stretching of the
cell faces, which is described by an additional linear term. The equation for the
theoretical elastic modulus of a closed-cell foam is expressed as [1]
E=Es C1 (f
r)2 S C2 (1 s f)
r
(1)
where Es is the elastic modulus of the solid material of struts and cell faces. f is the
volume fraction of solid contained in the cell edges, the remaining fraction (1 f) is
in the cell faces. The uniaxial compression (plateau) strength, s, of low-density opencell foams is scaled with r3/2. At higher relative densities the influence of the thickness of the struts and the size of the corner becomes considerable [2].
281
282
As shown in Fig. 6.2-1b the correction of the scaling law at medium densities
has a significant effect on the prediction of the value of plateau strength. The
theoretical value of plateau strength is modified for a closed-cell foam with the
stretching effect of cell faces as
r)3=2 S C4 (1 s f)
r
s=s ys C3 (f
(2)
where s ys is the yield stress of solid material and C1, ... C4 are constants depending
on the cell geometry [1]. These equations describe the behavior of a material that is
assumed to be continuous and homogeneous. In this study, our interest lies in
local variations of density and other dependent fields over scales not very much larger than the typical microscale. This may be accomplished with the help of a mesoscale window (averaging volume), which becomes the classical representative volume element (RVE) in the infinite medium. Mechanical properties at such a mesoscale can not be uniquely approximated. But they depend on the essential or natural boundary conditions of the averaging volume, which bounds the material response from above and below [3]. This bound will be tighter with increasing size of
mesoscale or, in other words, with decreasing effect of boundary conditions on the
response of the averaging volume. In the mesoscale, the relative density can be defined as a function of a local point X in an undeformed stress-free reference configuration, r r(X). This allows us to define an inhomogeneous mass distribution
in the foamed samples. Studies on the effects of inhomogeneous or gradient
distributions of mass on the behavior of metallic foams are limited. Beals and
Thomson reported that density gradients have a significant effect on compression
properties of Alcan aluminum foams [4]. Huschka et al. modeled foam materials
with a stacking of layers of different densities [5]. Gradinger and Rammerstorfer
developed a 1D model to analyze the effect of density gradient in loading direction
[6]. The effect of 1D inhomogeneities on the stress strain response and on the
crush energy absorption of aluminum foams has been studied. Daxner et al.
improved this method to a 2D model [7]. In this model, the material is re-
Figure 6.2-1.
strength.
0.05
/ ys
0.04
0.03
Alulight cast
Alulight wrought alloy
Alporas
0.02
Scaling law
=1
0.01
0
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Figure 6.2-2. Comparison of measured and predicted plateau strength of Alporas and Alulight
samples. The plateau strengths have been normalized by the yield stress of the cell wall.
283
284
6.2.2
3D Mesomodel
The density mapping method was used to approximate the cellular structure by a
continuous, 3D density distribution at the mesolevel. The density recorded by
XCT is homogenized over an averaging volume. The dimension of the averaging
volume specifies the size of mesoscale where the cellular structure is assumed
to be homogeneous. Overlapping of averaging volumes allows to increase the resolution of continuous density distribution up to the resolution of used XCT data.
Details of the density mapping method have been explained in Chapter 4.3 and
previously [8,10]. The influence of an averaging volume size (AVS) on the result
of the developed mesomodel will be discussed in Section 6.2.4.
The obtained inhomogeneous continuum body is then implemented by FE analysis. The mean density of each element is obtained by averaging the values of density mapping for all XCT voxels within the FE. Figure 6.2.-3. shows 2D XCT pixels
overlapping with a triangular FE. The next step is to describe the inhomogeneous
body as a multiple sub-domain problem. The material properties (elastic modulus
and Poisson's ratio in the elastic regime and plateau strength in the plastic regime)
depend on the local relative density of each element. In a general case, each element will have a mechanical behavior that is different from the behavior of neighboring elements. This high variation in the material properties can complicate the
analysis of the model. To overcome this difficulty the relative density variation,
[
rmin, rmax], is divided into N intervals. All elements, in which the relative densities
lie in the interval i (
rFE [
ri, ri 1]) form a sub-domain Vi (i 1,2,....,N). Therefore
the material consists of N sub-domains each of which is assumed to be homogeneous and isotropic. Their mechanical properties depend on the average value of
the relative densities of the elements lying within the sub-domain.
Although the mechanical properties in each sub-domain are assumed to be constant, they show discontinuities across any interface between sub-domains. The interfacial conditions can lead to deviation from uniaxiality of the loading condition
in the sub-domains even in the case of uniaxial compression of samples. Because
of this, the general form of constitutive laws is used for the description of elastic
and plastic behavior of the sub-domains.
Figure 6.2-3.
Elastic Regime
The generalized Hooke's law is used to calculate the linear elastic strain tensor in
the form
6.2.2.1
eei Sij s j
i, j 1, 2, . . . 6
(3)
where eie, s j are the elastic strain and stress tensor, respectively, and Sij is the compliance matrix. This matrix is characterized by a set of moduli. Two independent
moduli describe an isotropic and homogeneous material. The Young's modulus
and the Poisson's ratio are used for this purpose. The Young's modulus of each
sub-domain is derived from the scaling law (Eq. 1) with the assumption that the
material behaves like a closed-cell foam. The volume fraction, f, should be obtained by fitting the experimental results.
The Poisson's ratio is assumed to be 0.35 in the elastic regime. There is no experimental evidence for this parameter but the computational modeling of a regular cellular structure (Kelvin cells) suggests 0.35 as a reliable value [15].
Plastic Regime
The linear elasticity is limited by the yield criterion. Plastic yielding is the most
usual mechanism for failure of ductile metals. This mechanism can be described
by rate-dependent incremental plasticity theory. According to this theory, the elastic
theory breaks down as a certain function of stress reaches a certain value. It is
known as a yield surface within which the material is elastic and on which the plastic flow may take place. Various models have been developed to describe the shape
of a yield surface for a homogeneous cellular material [16 18]. The present investigation focuses on three models: classical J2 plasticity [19], the crushable foam
model [20], and the simplified self-similar model [16].
6.2.2.2
Classical J2 Plasticity
The classical J2 plasticity is the simplest and most common model to predict the
plastic deformation in dense metals. The yield surface in this model is expressed as
f se s Y 0
(4)
where Y is the yield strength of the material under uniaxial load and s e is the von
Mises effective stress defined by
r
3
sij sij
se
(5)
2
sij is the deviatoric stress tensor. This yield surface is independent of the mean
stress. Such a material can not fail due to hydrostatic load and is plastically incompressible (epvol) when an associate flow rule is assumed [19]. Because of plastic compressibility of cellular structure, the J2 plasticity is used to estimate the mechanical
behavior of foams at small global strain values and to determine the position where
local deformation begins to form a band. Strictly, J2 plasticity is not valid for
285
286
pt s pc
Sp
2
!2
S
t
M
!2 #1=2
s
pt S pc
0
2
(6a)
where
"
1
1
s
t se 1 S
2
K
1
1s
K
r
se
!3 #
(6b)
where pc and pt are the strength of material in hydrostatic compression and tension, respectively. M is the slope of critical state line that depends on pt and on the
initial yield strength under uniaxial load, Y, as well as under initial hydrostatic compression strength, pc0. K is a material parameter that defines the shape of yield surface
in the deviatoric plane (Fig. 6.2-4a), and r is the third deviatoric invariant of sij.
The yield surface is depicted in Fig. 6.2-4b. Non-associated flow is assumed for
this constitutive model. The flow potential, g, is defined in term of hydrostatic and
deviatoric effective stress, and chosen in this model as
r
9 2
p S s 2e
g
(7)
2
This potential gives the zero plastic Poisson's ratio. This means that loading in
any direction causes insignificant deformation in the other directions perpendicular to the loading. Experiments don't always confirm this property for foamed metals, specially those that have high relative densities [16]. This is a limit of the
crushable foam model to describe the plastic deformation of foamed metals. Another restriction is the calibration of the model with the experimental results.
K=1.0
hardened surface
K=0.8
pt
S3
plane.
pc|0
pc p
original surface
(b)
(a)
Figure 6.2-4.
softened surface
S2
Here, the hardening/softening response is controlled by the value of the hydrostatic compression strength pc. In other words, compaction or dilation of the material, which is modeled by variation of the yield surface's shape as shown in
Fig. 6.2-4b, is calibrated with the response of the material in a hydrostatic compression test. It is assumed that pt remains fixed throughout any plastic deformation.
Generally, we define the pressure pc versus total plastic volume strain epl
vol as
follows.
a) plateau regime
pl
pc k1 evol S pc0
pl
(8a)
pl
(8b)
evol J eD
b) densification regime
pl
pc k2 eaevol
eD J evol
k1, k2, and a are material parameters that depend only on relative density. They
must be evaluated by fitting the p e curves obtained by hydrostatic compression
tests. Parameter eD (densification strain) is adjusted to satisfy the continuity of
the function pc(epl
vol) in the passage from zone (a) to zone (b).
Simplified Self-Similar Model
The model was developed by Deshpande and Fleck [16] assuming the hardening
response is given directly by uniaxial compression responses. Hence the model
can exactly predict the compression response of a homogeneous material that is
loaded uniaxially. Multiaxial loading causes disagreement of the model predictions
with experimental results.
In uniaxial loading of an inhomogeneous material, the interface forces cause a
multiaxial loading condition in the sub-domains. This can also lead to deviation
of material behavior from the prediction of the model. This deviation depends
on the order of inhomogeneity (difference between lowest and highest density)
as well as the shape of sub-domains in the foamed sample. This constitutive law
is employed for modeling the uniaxial compression (including plateau and densification regimes) of real foamed samples as an inhomogeneous material. For this
287
288
pl
etrue J eD
true
(9a)
b) densification regime
pl
s true cedetrue
pl
eD
true J etrue
(9b)
c s es detrue
(10)
6.2.3
Uniaxial compression tests are simulated with respect to the presented model. The
compression samples are rectangular prisms with dimensions of 20 q 20 q
40 mm3. ABAQUS/Standard is employed for computation of the field variable (displacement, strain, and stress). 3D solid (continuum) elements with eight nodes of
type C3D8 are used to model the sub-domains. The nodes of this element have
three translational DOFs (degrees of freedom) but no rotational DOFs. The boundary conditions are chosen so that they represent the conditions of uniaxial compression tests. For this purpose, one clamping surface (bottom face) is assumed
to be fully fixed within the clamping plane, so all nodes are fixed in this plane
(x-, y-, z- translational displacements are assumed to be zero). Note that because
of the friction effect, in practice, the lateral displacement of a sample on the clamping surface is also negligible. Nodes in the other clamping surface (top face) are
fixed in the lateral directions (x and y) whereas the displacement in the axial direction (z) is given as loading.
Deformation Band
The compression experiments indicate that the inhomogeneous density distribution leads to localization of deformation and the formation of deformation
band(s) [11,12]. Figure 6.2-5 depicts the lateral surfaces of an Alulight sample recorded by a digital camera at 0 and 4 % nominal strain. The sample has an average
density of 350 kg/m3. The deformed cells are marked by arrows. It can be seen that
the deformed cells form a band in the sample. The edge length of the averaging
volume chosen to homogenize the density surrounding a XCT voxel was 6.4 mm
(11 pixel). The edge length of elements is chosen to be three times the size of
the CT pixel in each slice.
6.2.3.1
cells are marked with white arrows; c) calculated local plastic strains in loading direction
using presented model at 4 % nominal strain,
yielding the first deformation band.
289
290
Figure 6.2-6. Results of deformation modeling using the crushable foam model for densification
of first deformation band and forming the second band.
Figure 6.2-7. Simulation of large deformation of an Alulight sample using the simplified selfsimilar model.
291
292
To compute the effect of inhomogeneity the investigated sample was divided into
nine sub-domains. The densities of the sub-domains were 220, 240, 260, 280, 330,
400, 500, 620, and 720 kg/m3. The calculated plastic strains in the loading direction
happening at a nominal strain of 4 % are depicted in the Fig. 6.2-5c.
Very good agreement is seen between the position of the observed and the computed deformation band. Densification of this band and forming the second deformation band was modeled using the crushable foam model as shown in Fig. 6.2-6.
Large deformation of another Alulight specimen was simulated using the simplified self-similar model as shown in Fig. 6.2-7. It should be noted that field parameters (such as stress, strain, and displacement) are computed for a continuum.
They are average values and must not be considered as microscopical field parameters related to deformation of the cell walls.
Mechanical Properties
Figure 6.2-8 shows the predicted stress strain curve of a typical Alporas specimen.
The simplified self-similar model described above was used to model the hardening response during uniaxial loading. Although no strain hardening was assumed
for sub-domains in the plateau regime, the response of the whole specimens shows
hardening, which follows the measured response accurately. This phenomenon
can be described by large deformation of a sub-domain or of some elements in
a sub-domain, which can cause local densification. Rapidly increasing stress in
densification regime for these elements can lead to slow hardening of the whole
specimen in the plateau regime. With an increasing number of elements that
are densified the slope of stress strain curve increases quickly. The material parameters in the Eq. 9 are obtained by fitting the measured data and are considered in
this simulation as follows [21]:
6.2.3.2
Figure 6.2-8. Comparison of the measured stress strain curve of a typical Alporas sample with
the prediction of the mesomodel.
s 0:507r1:56, d 2:0, eD
true 1:0 s 2:13 r
(11)
0.08
0.07
E/Es
0.06
0.05
0.04
0.03
Exp.
mesomodel
Phi=1
Phi=0.9
Phi=0.94
0.02
0.01
(a)
0
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
0.24
0.26
0.06
0.05
p / ys
0.04
0.03
0.02
Exp.
mesomodel
Phi=1
Phi=0.94
Phi=0.94 (correc.)
0.01
(b)
0
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
0.24
0.26
Figure 6.2-9. Comparison of measured values with predication of the mesomodel based on the
simplified self-similar model: a) elastic modulus; b) plateau strength.
293
294
can be seen that this effect on elastic modulus is small, while the plateau strength
can be significantly affected from inhomogeneous mass distribution. However, the
overall response of foamed samples depends on the size and shape of soft (sub-domains with low densities) and hard (sub-domains with high densities) regions. The
plateau strength will be decreased specially when a soft region is oriented perpendicular to the external compressive load [10,12].
6.2.4
Discussion
3.5
Stress [MPa]
2.5
1.5
1
Exp.
AVS=11 pixel
AVS=23 pixel
ADS=33 pixel
0.5
(b)
0
0
Figure 6.2-10.
295
296
Figure 6.2-11. Influence of averaging volume size (AVS) on the density distribution derived from
XCT data (0.588 mm/pixel).
6.2.5
Conclusions
A mesomodel has been developed to explain the localized deformation behavior and
to calculate the mechanical properties on the basis of the mass distribution of the
cellular metals measured by CT data. Density mapping method is employed to approximate the mesoscopical continuous distribution of density in investigated samples. The model confirms the correlation between the inhomogeneous density distribution and the localization of plastic strains yielding deformation bands. The position of forming and propagation of the deformation band is predicted accurately.
The influence of inhomogeneity on the elastic modulus (the stiffness of the cellular structure) and yield strength of metallic foams can be demonstrated. The influence on the elastic modulus is small while the plateau strength can be affected
significantly. The discrepancy between the calculated and the measured plateau
strength is caused by the uncertainties for the parameters used for the scaling
laws for Alulight and effects due to the assumption of isotropy.
References
1. L. J. Gibson, M. F. Ashby, Cellular Solids:
Structure and Properties, 2nd edn, Cambridge
University Press, UK 1997, Chapter 5.
2. L. J. Gibson, M. F. Ashby, Proc. R. Soc. Lond.
1982, A382, 43 59
3. M. Ostoja-starzewski, Int. J. Sol. Struct. 1998,
35, 2429 2455.
4. J. T. Beals, M. S. Thomson, J. Mater. Sci.
1997, 32, 3595 3600.
5. S. Huschka, S. Hicken, F. J. Arendts, in Proc.
Metallschaume, J. Banhart (ed.), MIT, Bremen
1997, p. 189197.
6. R. Gradinger, F. G. Rammerstorfer, Acta
Mater. 1999, 47, 143 148.
7. T. Daxner, H. J. Bohm, F. G. Rammerstorfer,
Comput. Mater. Sci. 1999, 16, 61 91.
8. H. P. Degischer, A. Kottar, in Metal Foams
and Porous Metal Structures, J. Banhart, M. F.
Ashby, N. A. Fleck (eds), MIT Verlag, Bremen
1999, p. 213 220.
9. A. Kottar, H. P. Degischer, B. Kriszt, Materwiss. Werkstofftechn. 2000, 6, 465 469.
10. B. Kriszt, B. Foroughi, K. Faure, H. P.
Degischer, in Metal Foams and Porous Metal
Structures, J. Banhart, M. F. Ashby, N. A. Fleck
(eds), MIT Verlag, Bremen 1999, p. 241 246.
11. B. Kriszt, B. Foroughi, A. Kottar, H. P.
Degischer, Proc. Euromat 99, Vol. 5, WileyVCH, Weinheim, 2000, p. 4 82.
297
7
Service Properties and Exploitability
7.1
C. Haberling
Lightweight construction is gaining importance in automobile technology. Alongside improvements in wind resistance and tribology, weight reduction measures
are of primary importance. This is necessary because for each particular automobile class there are increasing customer demands for safety, comfort, performance,
and available internal space. Using conventional methods of construction to provide these customer-oriented features leads to a considerable increase in weight,
giving rise to strong pressures to reverse the weight gain. It is therefore regarded
as an imperative development objective to achieve a reduction in weight relative to
axle loads.
7.1.1
Introduction
299
300
Plate structures:
p
3
E Z flexual stiffness
r
(plate)
Rp0,2
Z tensile strength
r
p
E
Z bending stiffness
r
(bar)
Hollow structures:
q
3
Rp0,22
Z bending strength
r
(bar)
Rp0,2
r
Figure 7.1-1.
forcements.
Z tensile strength
Typical loading of car body structure (Audi A8): strategy for structural foam rein-
7.1.2
The areas where these kinds of material are of potential use can be derived at an
early stage in design using a finite element (FE) analysis of typical stresses in a car
body. The result makes it possible to evaluate which structural components in the
body structure can play a multifunctional role and what measures can be adopted
to achieve the desired result cost-effectively, taking into account space availability
and the factors inherent in lightweight construction. As illustrated in the car
body of the Audi A8 (Fig. 7.1-1), there are many areas in the bodywork where structural foams can be used. They are particularly suitable for hollow structures where
space is too restricted for other more cost-effective measures. Particular areas
where the multifunctional applications profile can be exploited to beneficial effect
are the side-members, the boundaries of the floor of the passenger compartment,
and in the region of the sills and pillars.
7.1.3
Material Properties
The application profile of a structural foam can be derived from the density of the
material, its compressive strength, and its elastic modulus. These are illustrated
(Fig. 7.1-2) for a single construction method. It is quite possible to adapt the behavior of the material specifically to the local stress relationships within the structure
of the vehicle owing to the wide range of material densities, types of material, and
Figure 7.1-2.
heat treatment methods that are available. Typically, metallic foams have a higher
elastic modulus, whereas epoxy foams exhibit a greater compressive strength at
similar densities for temperatures up to 80 hC.
7.1.4
Using lightweight construction criteria [3], it is possible to use these material characteristics to assess which type of material is most suited for what kind of stress
conditions (Fig. 7.1-1). Compared with compact materials, however, the efficiency
of foam materials in single construction methods where the available building
space is not an issue is somewhat limited. It is only where a differential method
of construction is used with compact materials and in limited spaces, which is
the predominant case encountered in automobile construction, where the effectiveness of these types of foam materials comes into its own. Even greater advantages
are gained when they are used in conjunction with compact materials. In the first
place, all foam materials exhibit an extremely variable behavior under tensile and
compressive loads. While the compressive range is characterized by a broad, uniform, and even deformation behavior (plateau) under constant load, under tension,
structural foams inherently become brittle and fail at very low loads (Fig. 7.1-3). In
addition, inhomogeneous regions are formed in the material that do not particularly affect the overall behavior under compressive stress (reaching the ultimate
strength either later or at a lower level) under purely axial loads, but may cause
buckling or failure of the foam structure if there is an uneven distribution across
301
302
Figure 7.1-3.
Figure 7.1-4.
the stressed cross section, and thus lead to a premature total failure (Fig. 7.1-4).
This would represent a risk particularly in the case of the complex stress patterns
in a car body that can certainly be avoided by using a carrier structure.
The foam materials on the market today that are used for structural stiffening
are ideally used inside hollow components or inside a sandwich composite structure (Fig. 7.1-5). The function of these carrier materials is to avoid high local tensile stresses in the foam and to apply the forces acting externally into the core of
the foam, directing them during deformation. It is of primary importance, particularly in circumstances where energy absorption is required, to obtain the correct
geometric balance in the carrier/foam system. A completely foamed tubular structure that is subject to bending can serve as an example. While the empty tube de-
forms considerably under a low load, when it is filled completely with foam, its
flexural resistance is increased considerably, but, at the same time, it will fail earlier
with smaller deformations. This can be attributed to excessive local stresses in the
outer tube structure that is overloaded owing to the restricted plastic deformation
locally, causing brittle fracture. It is more advisable to provide a partially foam-filled
structure where material is saved in the region of the neutral axis or inside the less
active tensile zones in the case of structural foams. In this way, weight can be
Figure 7.1-5.
Figure 7.1-6.
303
304
In general, there are many ways of applying and attaching structural foams. The
effectiveness depends essentially on the attachment of the foam materials to the
surrounding or adjacent components. This is particularly important in structures
that derive their properties from a composite sandwich structure. Frequently, however, these requirements conflict with production-related conditions such as manufacturing tolerances. Several additional factors affect the stiffening of hollow
bodywork structures in car construction. In one such instance, it must be ensured
that the process liquids in the paint shop flow through the hollow cross sections of
the bodywork, so that, in general, this objective is defeated if larger hollow sections
are completely filled with foam. To prevent corrosion zones, narrow gaps must be
avoided between foam elements and the basic structure, particularly where mixed
construction methods are used, such as the employment of aluminum foams in a
steel body. Also, incomplete contact must be avoided for acoustic reasons. The contact must be such that either it is so intimate that contact is permanently assured
over the entire surface, or both parts must be protected separately against corrosion
before assembly, for example by painting. However, this solution generally incurs
considerable extra costs. In the case of cellular metals, both cementing and casting
are feasible means of bonding surfaces. Local joining can be done using thermal
and mechanical bonding processes.
Casting
The casting of aluminum foams using sand and permanent molds can be regarded
as state of the art. Cast-in molding in pressure die-casting is possible using diecasting machines controlled in real-time to avoid pressures in the molten metal becoming too high. Most die-cast parts used in car construction have a typical wall
thickness of less than 5 mm. In order to deal with all aspects of lightweight construction in this range of applications, it is necessary to explore at what maximum
pressures and casting speeds aluminum foam inserts of minimum density can be
cast-in with the thinnest possible casting skin while avoiding both foam infiltration
and incomplete filling of the mold (Fig. 7.1-7). This is particularly important for
components with large hollow sections (such as swivel bearings) that are normally
produced by gravity molding with a lost-sand core (see Section 3.3).
Apart from casting, gluing is a process that can be used to join together the surfaces of foam parts. In car production it is especially important to achieve a sufficient adherence and durability against corrosion and aging. Expanding adhesives
can be used to compensate for even greater manufacturing tolerances, so that, for
instance, it is possible to insert a foam part in an extruded section (Fig. 7.1-7)
(see Section 3.2).
7.1.5.1
Figure 7.1-7.
Figure 7.1-8.
305
306
ances of the parts being joined. Other welding methods require appropriate preparation of the carrier system, such as localized perforations, since the depth of
the weld is shallow. Welding with and without filler materials has been checked
in numerous ways. When employed in fabricating car bodies, there still remain
questions regarding the corrosion behavior in narrow gaps along the weld seam.
Furthermore, the positions of the seams must be adjusted to match the stress profile of the component. This is necessary to ensure the maximum effectiveness of
the foam inlay, and, at the same time, to minimize as much as possible the weakness of the surrounding sheet metal along the seam, since typically in aluminum
welds the zone around the weld is subject to softening.
Three-dimensional sandwich
A further solution is offered by the use of a prefabricated three dimensional sandwich
structure. The most notable example of this type of construction is the aluminum
sandwich foam (AFS) sheet [4], in which the joint between the foam core and the
outer plate is produced in a roll-bonding cladding process before foaming in situ
(see Section 3.4). Since the foaming process does not take place until after the conversion process at temperatures over 500 hC, 3D sandwich contours can also be fabricated using this material. These sandwich structures are either screwed or welded
to a surrounding aluminum structure using the cover plates.
7.1.5.4
Figure 7.1-9.
7.1.6
Effectiveness
Physical characteristics can be checked using simple test pieces for bending, impact, torsional, and axial stresses that mirror the geometry of how hollow structures are used in a vehicle. In doing so, the assumption is made that the outer
dimensions of the construction space that is being optimized are not allowed to
change since an increase in the cross section of the carrier profile generally represents the most effective lightweight construction solution. This practical necessity
occurs frequently in vehicle production owing to space restrictions. The test piece
selected was a closed spot-welded hat section made from DC-04 sheet steel, filled
with U- and O-shaped foam pads with a density r 0.55 g/cm3 made from aluminum or epoxy foam. The properties evaluated were stiffness under bending,
or flexural strength, impact strength behavior, torsional stiffness, and torsional
strength. This was compared with a section of the same weight with thickened
walls (Fig. 7.1-9).
307
308
Figure 7.1-10.
creased by using an O-shaped structural foam insert [5]. Nevertheless, the effectiveness is considerably less than when a bending stress is applied. Removing the effect of weight (t max,elast/m), reinforcement using 4 mm thick epoxy foam produces a
20 % improvement compared with an initial section without the insert. Using an
8 mm thick aluminum foam pad (this reflects the minimum wall thickness currently possible under production conditions when producing more complex
parts) results in a torsional stiffness that is scarcely any different than if one
used a 4 mm thick epoxy foam of the same density, with adhesion across the
complete surface. This is despite the higher elastic modulus and the greater
wall thickness, and it occurs because the surfaces for gluing the upper and
lower sides are reduced for technical reasons. In this case, therefore, owing to
the total additional weight, the aluminum foam does not offer an advantage
for lightweight construction. This applies also to unreinforced sections with
increased wall thicknesses used for this application, where the increased weight
make them unsuitable for lightweight construction since, again, they offer no
advantages.
Impact Stresses
The obvious advantage offered by a structural foam insert in relation to impact
strength is the strengthening and energy absorption that it confers on a hollow
structure even where there are large dynamic bending deformations. The behavior
can be modeled with great precision in an FE simulation. Calculations and tests on
a hat section show that, at an impact speed of 3 m/sec using a mass of 600 kg on a
section reinforced with a U-shaped structural foam insert joined to the surface of
7.1.6.2
Figure 7.1-11.
the external plate, the failure load and the energy absorption are more than
doubled compared with an unreinforced hollow section of the same weight and
with thickened walls. This applies also to epoxy foam variants and for aluminum
foam elements glued on all sides. If the same aluminum foam element is glued
only locally, however, such as along the cover plate, the stiffening effect is reduced
considerably (Fig. 7.1-11). If the energy absorption is spread through the body of
the composite component, the same pattern emerges. However, it is also evident
that the effectiveness in relation to the mass used falls off again when the foam
thicknesses are too high.
Axial Load
Tests have been carried out on different sections in many different ways to study
their effectiveness under axial loads with and without foaming in place. The buckling associated with folding that is usually used for this type of loading applied to
hollow structural parts can also be used in cases where structural foams provide
stiffening and can bring about significant improvements in its efficiency [6]. A decisive element in this is a reduction of the force levels as the folding is taking place.
Nevertheless this improvement is largely offset in many instances when one takes
the added weight into account [7]. Furthermore, the effectiveness of the empty
structure can be improved significantly by making certain changes in shape or
by using alternative composite fiber structures (carbon/aramid fibers in an epoxy
matrix) (Fig. 7.1-12) with little increase in weight [2].
7.1.6.3
309
310
Figure 7.1-12.
Acoustics
Acoustic effectiveness has been tested on different models (see also Section 3.4.3.1).
Large AFS panels display good absorption qualities below 500 Hz [4]. Other compact foam parts also provide good absorption characteristics but only within a limited frequency range. Tests performed on foam-filled engine supports (Fig. 7.1-13)
7.1.6.4
A: diecast
engine support
B: aluminium
foam-insert
20 mm
n
m
C: combination
of A and B
Figure 7.1-13.
D: alternative
solution
Aluminum-foam filled engine support for acoustic means and alternative solution.
revealed high acoustic effectiveness, but were considerably heavier than alternative
solutions using organic materials since the minimum thickness of the foam was
8 mm. At present, therefore, the acoustic efficiency provided by closed pore cellular
metals when used in car construction can only be regarded as an added benefit
when taken in the context of a multifunctional role.
7.1.7
Outlook
lighter
materials
Starting
point
Structural
modification
I=bh /12
W=bh /6
311
312
On the other hand, structural foams based on plastics are now found in numerous applications in car manufacture and are used in the structural regions illustrated in (Fig. 7.1-1). Large-scale use of these materials is, however, prevented in the
current cost situation, although many of the additional costs associated with metallic foams, such as extra joining operations and corrosion protection do not apply in
the case of these materials. It is possible to use them when the space available excludes any other solutions, or where immediate additional stiffening measures
need to be taken, or when, as in the new Audi A4, additional costs or variants
can be avoided in another place.
AFS sandwich panels in prototypes have shown that there is a potential for using
them also [7]. Even for this material, however, series production has raised new
questions about joining in mixed construction methods. Furthermore, vehicle design must be adapted to the AFS sandwich in order to extract the maximum efficiency from this material.
First and foremost, the quality of the material has to be improved before cellular
metals can be used more in vehicle technology. It is already possible, however, to
use them inside sandwich structures or in conjunction with carrier materials. Development effort is important to produce an efficient cost-effective joining method,
as well as investigations on corrosion protection for using them inside steel structures that are subject to corrosive conditions. Furthermore it is essential to reduce
further the present minimum wall thicknesses and the manufacturing costs in
order to be able to compete with comparable polymer-based materials and other
approaches.
References
1. H. G. Haldenwanger, Konzeptioneller
Leichtbau mit neuen Werkstoffen: Verfahren,
ein Zielkonflikt bei der Verifizierung der
Entwicklungsprozekette im Karosseriebau,
in Proc. Aachener Kolloquium, Fahrzeug und
Motortechnik, RWTH Aachen, IKA, VKA,
4 6 Oct 1999.
2. H. G. Haldenwanger et al., Leichtbau mit
alternativen Werkstoffen und Verfahren Ingenieur-Werkstoffe 1999, 8(1).
3. B. Ludge, Funktionaler RohkarosserieLeichtbau, in Proc. Aachener Kolloquium,
Fahrzeug und Motortechnik, RWTH Aachen,
IKA, VKA, 4 6 Oct 1999.
4. H. W. Seelinger, Application Strategies for
Aluminum-Foam-Sandwich Parts (AFS) in
Metal Foams and Porous Metal Structures,
J. Banhart, M. F. Ashby, N. A. Fleck (eds),
MIT Verlag, Bremen 1999.
7.2
Functional Applications
J. Banhart
x
x
7.2.1
General Considerations
The first point is, in particular, crucial for any evaluation of applications for cellular
metallic materials. Many applications require that a medium, either liquid or gaseous, be able to pass through the cellular material. There may be a need for various
degrees of openness, ranging from very open for high rate fluid flow to
completely closed for load-bearing structural applications, and appropriate materials satisfying these conditions have to be found. Figure 7.2-1 shows which types
of porosity the various application fields require. Normally, a difference is made between whether an application is functional or structural, but there is considerable overlap between these two notions.
The question of from which metals or alloys a given type of cellular structure can
be manufactured is also important. Structural load-bearing parts have to be light
because otherwise they would be made from conventional massive metals or alloys.
Therefore, aluminum, magnesium, or titanium cellular or porous metals are preferred for such applications. For medical applications, titanium may be preferred
because of its compatibility with tissue. Stainless steel or titanium is required
for applications in which aggressive media are involved or high temperatures
occur.
Finally, processing and cost issues have to be considered. The technology must
be available to bring the selected cellular metal into the required shape and to incorporate it into the machine or vehicle where it has its function. A technology for
making cellular metal will be futile if the required component cannot be manufactured at a reasonable price.
In the following sections, applications for cellular metals are discussed that are
predominantly functional in the sense discussed. Traditional powder metallurgy
313
314
structural
load-bearing
components
bio-medical
implants
energy
absorbers
catalyst
supports
kind of
application
silencers
sound
absorbers
filters
bearings
heat
exchangers
functional
open
partially open
closed
type of porosity
Figure 7.2-1.
(PM) has created porous sintered metals for a wide range of applications [1 3]. It
is therefore not surprising to find very similar applications for the cellular metals
described in the present text, provided they have a certain degree of open porosity.
7.2.2
Biomedical Implants
Biomedical applications often have both a structural and a functional aspect, and
are therefore very challenging. Titanium or cobalt chromium alloys are used for
prostheses or dental implants because of their biocompatibility. To ensure ingrowth
of tissue, one usually produces a porous layer of the same or another biocompatible
material on the prosthesis by thermal spraying or other methods [4]. Alternatively,
one could use porous titanium or titanium foam for such applications and tailor
the density distribution to meet the required strength and moduli of such components. There is no unanimity about how implants should be designed to ensure
maximum durability and functionality. According to one opinion, the modulus
of, say, dental implants should match the modulus of the jaw bone. Knowing
the relationship between modulus and density of metallic foams, one could easily
manufacture implants with an appropriately adapted modulus, ensure biocompatibility, and stimulate bone ingrowth into the (open) porosity [5]. Strength
and design criteria have to be met as well. Magnesium foams could be used as
biodegradable implants that serve as a load-bearing structure as long as the bone
still grows but are gradually absorbed by the body in a later stage of recoalescence
[4].
7.2.3
There are two types of filters: filters holding back and separating solid particles or
fibers dispersed in a liquid (suspensions) or filters holding back solid or liquid particles dispersed in a gas (smoke or fog). Examples of the first type are filters for
cleaning recycled polymer melts, for removing yeast from beer, or for contaminated oil. The second type includes filtration of diesel fumes or water removal
in air lines. Important filter properties are fine filtration capacity, good particle retention, cleanability, mechanical properties, corrosion resistance, and cost. Some of
the cellular metals described in this text possess a combination of properties not
covered by the traditional PM materials and might therefore be considered as complementary to these, such as the materials described in Sections 2.1.2, 2.3 and 2.4.
7.2.4
Highly conductive foams based on copper or aluminum can be used as heat exchangers [6]. In this case open-cell structures are needed such as those described
in Sections 2.3 and 2.4. Heat can be removed from or added to gases or liquids
by letting them flow through the foam and cooling or heating the foam at the
same time. Owing to the open porosity, pressure drops can be minimized (see
Fig. 7.2-2). An example of such an application is a compact heat sink for cooling,
for example, in microelectronic devices with a high power dissipation density such
as computer chips or power electronic components. Nowadays, fin-pin arrays are
the standard solution in such cases. Metal foams or ordered cellular metals of
the type shown Fig. 7.2-3 can perform better if they are selected so that thermal
conductivity is kept as high as possible with their flow resistance maintained
low. These two requirements are contradictory. Therefore, an optimization problem
has to be solved [7,9,10]. The variables are the dimensions of the individual struts
and their spacing in the example given. A denser arrangement of metal naturally
gives rise to a better heat conduction but also increases pressure drop. The optimum determined is represented by a certain area in the plane of the diagram
spanned by the heat dissipation and the pressure drop, both expressed as dimensionless quantities, in Fig. 7.2-3.
Another application field for open-cellular materials is transpiration cooling. The
high surface area, low flow resistivity, and good thermal conductivity of some of the
materials used make them promising candidates for such purposes.
315
316
Figure 7.2-2. Pressure drop in partially open-porous cellular solids of the type shown in the
middle of Fig. 7.2-1 [8].
7.2.5
One of the oldest applications of porous PM materials is as self-lubricating bearings in which oil is stored in the interstices between particles and is allowed to
flow slowly out, thus replacing the used oil. Of course, some of the cellular materials described in this book could fulfil the same function but with the advantage of
having a higher storage volume than traditional PM parts. The application is not
limited to oil: water can be kept and slowly released for automatic humidity control. Perfume can be stored and allowed to evaporate slowly. Porous rolls can
hold and distribute water or adhesives to surfaces. Transport of the liquid can be
driven by capillary action alone or by excess pressure in the roll. Finally, very
open metallic structures can be used to store fluids at a constant and uniform
temperature, for example, in cryogenic conditions. Moreover, the foam can
reduce undesired movements of the liquid in partially filled tanks (anti-sloshing)
[10,12].
7.2.7
Porous materials can be used for controlling the flow of liquids and gases [1]. It is
known that PM flow restrictors are more reliable and accurate than conventional
micrometering valves. Because so many degrees of openness of cellular metals
are available, one could find tailored solutions for even more applications by appropriately selecting a cellular metal. Metal foams have already been used as flow
straighteners in wind tunnels [12] or flow distributors in valves [13].
317
318
Figure 7.2-4.
7.2.8
Silencers
Spargers
Battery Electrodes
Lead foams could serve as supports for the active material in lead acid batteries in
replacement of conventional lead gratings, thus allowing for constructing very
light electrodes [14]. The electrochemically active mass, a paste containing very
fine lead oxide powders, could be filled into the open voids of a lead foam
where it would contact the electrolyte (sulfuric acid). The lead foam acts as a highly
conductive lattice leading to a low internal resistance of the battery. The nickel
foams described in Section 2.4 are already used as electrodes in pasted rechargeable
NiCd or NiMeH batteries where weight savings and a higher energy density can be
achieved [15 17]. Porous PM materials with their extremely high surface area are
being used in fuel cells [1,15].
7.2.11
Electrochemical Applications
Nickel foams can be used as electrode material in electrochemical reactors. In filter-press electrodes, for example, a stack of isolated metal plates is used. The plates
are separated by a turbulence-promoting plastic mesh and insulating membranes.
If these meshes are replaced by sheets of cellular nickel with open channels (with
each sheet attached to one of these plates), one increases the electrode surface
while maintaining the turbulence promotion [18]. The reactors can be built
more compactly this way. Nickel foams can also be used to improve electrocatalytic
processes such as the electrooxidation of benzyl alcohol assisted by NiOOH, which
is electrogenerated on nickel anodes. Packed beds of nickel foams were shown to
improve the performance of such reactors [19].
7.2.12
Flame Arresters
Cellular metals with high thermal conductivities of the cell-wall material can be
used to stop flame propagation in combustible gases. Open-cell foams of the
type described in Section 2.4 have been shown to be capable of arresting flames
even when they were travelling at velocities up to 550 m/s. In practice, long
runs of pipes transporting combustible gases are protected close to possible
sources of ignition so that, if ignition does occur, the flame cannot accelerate to
high velocities [13].
7.2.13
Water Purification
Acoustic Control
319
320
foams have been studied for their suitability as impedance adapters for ultrasound
sources. Noise reduction functions are described in Section 5.3.
References
7.3
Machinery Applications
Cellular metals have existed as particularly innovative new materials for several
years [1 3]. Since the development of foaming technology, a number of prototypes
for use in the car and machine tool industries have been demonstrated [4 6].
This Section describes possible applications of foamed metals in mechanical engineering.
7.3.1
Parameters
Thermal Behavior
To evaluate the thermal characteristics of metal foams, the response of sandwich
sheets with a foamed aluminum core to thermal impacts was investigated. The
sheets (400 mm q 300 mm) were clamped on one side and warmed by a heating
mat placed on the top cover plate.
The room temperature was kept constant at 20 hC by a thermal cell. The sheet
temperature was recorded by two probes, one attached to each of the bottom
and top plates.
The thermal expansion was recorded by path sensors and was transformed into
the resulting linear expansion coefficient. The expansion coefficients of foamed
sheets covered with aluminum range from 0.020 to 0.024 mm/K mm. These values
are slightly higher than those of the AlSi12 alloy (0.0199 to 0.0206 mm/K mm) that
had been foamed. The face plate expansion coefficient is 0.0235 mm/K mm, so the
cover plates dominantly influence the total expansion. The lower expansion coefficient of steel (0.012 mm/K mm) also diminishes the resulting strain on the sandwich sheet (see Fig. 7.3-1). The expansion coefficients of the plates covered by
steel sheet range from 0.0105 to 0.014 mm/K mm.
7.3.1.1
0,030
0,025
x-direction
z-direction
0,020
0,015
0,010
0,005
0,000
Alu
1 mm
Alu
2 mm
Steel
1 mm
Steel
2 mm
Expansion coefficients as a function of cover plate material and plate thickness for
a two-face sandwich (density 0.65 g/cm3; sandwich 25 mm).
Figure 7.3-1.
321
322
Figure 7.3-2.
0.7 g/cm3).
Maximal forces for different connecting elements (diameter M6, Al foam density
shape. The joint length with the screws was always 20 mm, but the nuts and inserts had a joint length that ranges between 6 and 20 mm.
7.3.2
Examples of Application
Foamed Steel Pipes
To make stiff, light structures hollow profiles of various shapes and dimensions are
often used. To make such hollow sections very stiff, very large cross sections are
required, but then the walls tend to oscillate. Filling hollow sections, or pipes,
with metal foam is one possible way of influencing frequency and damping vibrations.
The following pipe profiles were used, being rectangular steel pipes of various
cross sections and lengths, with different wall thicknesses:
Height q Width q Wall thickness
160 q 80 q 4 mm
length: 2000 mm
160 q 80 q 3 mm
length: 1000 mm
160 q 80 q 4 mm
length: 1000 mm
160 q 80 q 5 mm
length: 1000 mm
120 q 60 q 4 mm
length: 1000 mm
80 q 40 q 4 mm
length: 1000 mm
The purpose of these investigations was to find out about damping as a result of
foaming of pipe profiles with aluminum. The influence of the foams on the frequency spectrum can also be clarified. The investigation was performed to find
out how far the natural frequencies are shifted by foaming, and to determine
the dominant frequency and the sequence of the natural frequencies. For the design engineer, it is important to know whether the first natural frequency causes
a locally limited transverse vibration or a bending vibration of the entire structure.
Reference values were determined by finite element (FE) analysis and measured
on the hollow profiles. To minimize the influence of the fixture, at the nodal points
(the Bessel points), the pipes were laid on two thin wires (see Fig. 7.3-3). This installation is more advantageous than a one-end fixture.
Vibrations were excited by hitting the pipe on a free end using an impulse hammer. The attenuation of the freely damped vibration was recorded by an accelera7.3.2.1
Figure 7.3-3.
323
324
tion sensor. From this curve, we obtain the logarithmic damping decrement and
consequently damping, D
!
^1
1
X
(1)
D ln
^z 2p(z s 1)
X
Natural Frequencies and Characteristic Types
The simulation runs showed that the frequency spectrum depends on all three
pipe geometry variables: cross section, length, and wall thickness. Consequently,
the first natural frequency of long pipes is not necessarily lower than for shorter
pipes. Assuming identical cross sections, pipes with thin walls are characterized
by lower natural frequencies (see Table 7.3-1).
The first natural frequencies are mostly bending vibrations in the x direction,
which is the direction of minimum area moment of inertia. In two pipes, the
first frequency is a local vibration of the pipe wall. This only occurs if the wall is
very thin or if the wall-thickness to cross-section ratio is small. Such wall vibrations
are almost excluded by the metal foam, since the transverse motions are hardly
hindered. For that reason, these frequencies are shifted towards higher values
(Table 7.3-2). For bending vibrations, frequency can only be improved if the stiffness grows more than the weight due to foam filling, but this cannot be achieved
for thick-walled pipes.
As shown, the first bending frequency for the hollow pipe appears as the 5th
frequency only. For the foam-filled pipe, wall vibration can totally be avoided and
only appears in conjunction with bending vibrations of the entire pipe at higher
frequencies.
The frequencies determined experimentally were in good agreement with the
simulation values, which is crucial for damping measurement. If the measured
Table 7.3-1 Natural frequencies of hollow profiles and foam-filled profiles. Vibration type: W
local oscillation of pipe walls; X oscillation in x direction (direction of minimum area moment
of inertia).
Pipe type
Cross-section
[mm]
Length
[mm]
Hollow profile
First natural
frequency [Hz]
Characteristic type
Foam filled
First natural
frequenzy [Hz]
Characteristic type
160 q 80 q 4
2000
151.3
134.4
160 q 80 q 3
1000
324.9
493.4
160 q 80 q 4
1000
432.7
506.3
160 q 80 q 5
1000
531.2
513.8
120 q 60 q 4
1000
430.4
397.1
80 q 40 q 4
1000
290.8
271.9
Table 7.3-2.
Frequency
no.
Hollow profile
Foam filled
Frequency
[Hz]
Characteristic type
Frequency
[Hz]
Characteristic type
325.0
Walls oscillate
493.4
First bending in x
369.6
Walls oscillate
805.7
First bending in y
459.2
Walls oscillate
1135.0
First torsion
476.6
Walls oscillate
1222.5
Second bending in x
479.0
First bending in x
1879.5
Second bending in y
530.8
Walls oscillate
2102.5
Second bending in x
whereby the walls are
slightly buckled
damping values are to be integrated into the simulation runs, it is of particular importance that the calculated and the measured values coincide. This is the only way
making sure that the measured damping values can be used for the calculation of
component dimensions
Damping of Structure
Damping was determined at the same time as the frequency was measured. Contrarily to expectations, the location of the supports did not influence the frequency
and the damping parameters. This is because, in a dynamic understanding, the
ropes used are soft. So the ropes would only influence a much lower frequency
range than that investigated here.
Having kept all the other conditions constant, even the scanning rate was without any obvious influence: when comparing scanning procedures at 2.5 and
1.25 kHz, the difference between both damping values was merely 0.46 %.
The evaluation showed that the damping factor was slightly higher for the first
vibrations than for more attenuated vibrations. In one experiment, damping between peak 200 and 700 was D 0.66 q 10 5, and between 1000 and 1500, it
was D 0.64 q 10 5. This effect is caused by the higher damping influence of
the surrounding air for high amplitudes. To minimize this influence of about
3 %, the evaluation range was selected as widely as possible.
In the follow-up investigation of pipes, the damping values shown in Fig. 7.3-4
were determined for the corresponding frequencies. In the investigation, the dependency of damping on the pipe cross sections or the wall thickness could not
be recognized. Instead the damping values are related to the natural frequency.
It makes sense to compare the damping of equal pipe frequencies before and
after foaming.
325
326
800
700
Tube 1
600
Tube 2
500
400
300
200
100
0
1
10
11
12
13
14
15
16
17
Test no.
Figure 7.3-4.
Damping growth of the individual pipes (at 300 % change in mechanics assumed).
The increase in damping by foaming varied: the mean change ranged from 133
to 645 %, or 799 % for actual, rather than mean, values. Figure 7.3-4 shows that the
majority of measured values are located below 300 % (see line); few values lie
above, but these are notable. We suppose that the damping mechanisms that
acted in both types of pipes (above and below the 300 % line) are different. For
the types below the line, material damping is the active mechanism, assuming
that the foamed cores and the steel pipes do not move relative to each other,
thereby excluding any frictional or joint damping. These conditions are obtained
for an optimal foaming procedure with clean steel pipes, so a direct metal connection between the steel pipe and the foam core is brought about, without any gap
that would permit a relative motion.
If the steel pipes are slightly rusty, then metal-to-metal contact between the pipe
and the foam core may be incomplete, and some frictional damping may then
occur. Frictional damping depends on many factors such as pressure, surface
roughness, the medium in the gap, and others. In pipes lying above the 300 %
limit, a combination of material and frictional vibration can be assumed.
We cut sample sections from the foamed pipes, which confirmed that in some
there was good metal-to-metal contact and in others there was a visible gap
between the foam and the pipe wall. It is too simple to classify the foamed
pipes into good and poor categories, but the following statements can be postulated.
x
For foam-filled pipes in which the foam adheres to the inner steel wall in an optimum manner, damping is increased by up to 300 % in comparison with hollow
pipes.
Damping beyond this limit suggests that the foam has not been completely
bonded. This phenomenon could be used as the basis of a nondestructive quality
inspection of the foaming process.
Comparing the measured damping values with those taken from literature, we
note that a damping for steel of D 3.7 q 10 4 [7] is in the range of our steel
pipes. The values for metal foam range from D 9 q 10 4 to 2.1 q 10 3 and
are clearly higher than those of steel. Taking both materials together, a
higher total damping should be provided. However, a direct comparison is not
possible, since damping does not only depend on the material and load, but also
the shape of the specimen [8]. For that reason, the damping measured on a workpiece is sometimes defined as the workpiece vibration instead of the material
vibration.
Machine Table
On the occasion of the CPK `98 conference, the first functioning assembly filled
with metal foam was demonstrated. The part size was 800 mm q 720 mm q
250 mm, so it can be justifiably called an assembly relevant for practice. A conventional machine table (Fig. 7.3-5) on the linear motor test bench of the IWU in
Chemnitz was replaced a steel framework filled with aluminum foam, optimized
by FE analysis. The static and dynamic properties of the conventional assembly,
the steel framework, and the foam-filled framework were measured (Fig. 7.3-6).
The main goal was to reduce weight whilst maintaining stiffness and strength.
The weight of the foamed structure was 72 % of the conventional table (Fig. 7.3-7),
but deflection only decreased by 19 %. These results were obtained by increasing
the table thickness and having a higher area moment of inertia. The static behavior
of the foamed table is much better than the hollow steel frame, because of the improved stiffness and hence minimal deflection (Fig. 7.3-8).
7.3.2.2
Figure 7.3-5.
Figure 7.3-6.
327
Figure 7.3-7.
Comparison of weights.
Figure 7.3-8.
Comparison of maximum
Figure 7.3-9.
Mass [kg]
400
Solid table
300
Stell frame
200
Foamed table
100
0
Deflection [m]
25
20
15
Solid table
Steel frame
10
Foamed table
deflection.
0
600
500
400
Solid table
300
Steel frame
200
Foamed table
100
teristic type.
Very good results were obtained for dynamic parameters: the first natural frequency was increased by 80 % (Fig. 7.3-9), which is mostly due to wall vibrations
being almost eliminated by the foam. Such vibrations can only be detected at very
high frequencies.
The damping was multiply increased for the foamed version: for the first natural
bending frequency, a 10 times improvement was achieved (Fig. 7.3-10). Increased
damping arises from the material damping of the metal foam and from the properties of the adhesive that bonds the steel and the foamed parts.
This result is of special interest for applying the technique in mechanical engineering. In present machine tools, assemblies move with enormous accelerations.
To meet the increased demands for manufacturing accuracy, surface quality, and
tool life, the vibrations resulting from deceleration, acceleration, and the manufacturing process, have to be damped in an efficient manner.
2
Damping [%]
328
1,5
Solid table
1
0,5
0
Steel frame
Foamed table
Figure 7.3-10.
frequency.
Cross-Slide
As a second example, the cross-slide of a high-speed milling machine constructed
from steel and foamed aluminum is shown after fabrication (Fig. 7.3-11). The
cross-slide represents the connecting element between the z slide spindle and
the machine bed.
Obvious improvements compared with conventional castings can be anticipated
from simulations (Fig. 7.3-12). Thus, weight is reduced by 46 %. The deformation
deteriorated by 12 %. This worsening could be fully compensated by increasing the
wall thickness. Even then, an evident mass reduction would remain.
Nevertheless, the first natural frequency is increasing by 35 % in any case, which
means a clear enhancement in the field of mechanical engineering. Apart from
this, the position of the natural frequencies can be selected by varying the foam
density.
7.3.2.3
Figure 7.3-11.
140
Change [%]
120
100
80
60
40
20
112
135
54
0
Mass
Figure 7.3-12.
Total deformation
1. natural frequency
329
330
7.3.3
Conclusions
Metal foams are characterized by very useful physical and mechanical characteristics. Examples from mechanical engineering show some beneficial properties of
hollow components and sandwiches when filled the metal foam material and
how to make use of them in innovative products.
References
7.4
R. Kretz
When metallic foams were revived by new manufacturing technologies such as powder metallurgy, investment casting, and other processes in the early nineties, the
new material was introduced to the industry and to potential customers in a very enthusiastic manner and with great advertising expense. Information was widely
spread by companies and institutes who worked on the development of metallic
foams. Metallic foams became highlights at industrial fairs and at many conferences
researchers, designers, and materials engineers discussed the structures and the
properties of cellular metals that are relevant to industrial application. The foam developers missed no chance to present their work and to introduce their new product,
although the first material samples were only demonstrators: they were small, had
very irregular pore sizes and pore distributions, and a high degree of scattering in
their properties. They were all hand made and their producers were very proud
of their creations from porous metals. With the growing knowledge and experience,
manufacturing methods were further improved and very soon the presented foams
were of a respectable quality with a certain reproducibility. Whenever people held the
new material in their hands, they were impressed and almost everybody had a new
idea concerning an application of metal foams.
7.4.1
Introduction
But very soon the discussion came round to real parts or components and then the
foam manufacturers had to think about manufacturing net-shape parts, considering particular problems of geometry or function. Since real net-shape parts are not
flat plates, cubes, or cylinders, new manufacturing methods using mold and furnace technologies had to be introduced. The manufacturing of the powder metallurgical (PM) precursor material as well as the foaming technique for the compacted powder are called the powder compact foaming technique (PCF), which
is practiced by IFAM, Bremen, Germany, Leichtmetallkompetenzzentrum, Ranshofen (LKR), Austria, and the Slovak Academy of Science (SAS), Bratislava, Slova-
331
332
From the very beginning the application of the PCF route for making net-shape
parts was the goal. This method, using compacted aluminum powder and titanium
hydride as a foaming agent, is described by several authors [1,2].
x
x
x
x
x
x
333
334
Foaming Process
For successful manufacturing of aluminum-foam parts in the PCF, it is necessary to
observe furnace temperature, heating rate, temperature distribution within the
mold, foaming time, and cooling conditions that have to be kept in certain ranges
to achieve aluminum-foam components of high quality. These parameters are not
the same for all types of component; they depend on their geometry and dimensions
(wall thickness), on the alloy, and also on the properties and design of the mold.
In prototype manufacturing it is not possible to modify the whole equipment or
to change the layout of machines whenever a new component with a new geometry
has to be foamed. Compromises concerning the furnace used and the process parameters must be made to enable the best possible technical and economic manufacturing conditions to be achieved. Nevertheless, it is possible to make parts in a
wide range of different geometries with the use of adequate equipment. From
the beginning of aluminum-foam development, prototype manufacturing was
the main goal. This policy led to a large number of prototype projects producing
components with very different geometries and applications. The development of
the foaming process, applied to a particular component, was carried out along
with component testing to verify material properties.
In series production it would be necessary to adapt the furnace and the mold to
the product to achieve the optimized foaming parameters as well as the most economic production conditions for the particular product. The number of parts that
has to be produced is also an important process parameter to be considered in the
design of an aluminum-foam production machine and the necessary additional
equipment.
7.4.2.2
Foaming Furnaces
One of the most important pieces of equipment in the aluminum-foam manufacturing route is the foaming furnace. The foaming parameters have to be kept in
their optimized range, the precursor material in the mold has to be heated to
the foaming temperature in the foreseen time, and the temperature distribution
within the mold must be kept within small tolerances to avoid irregular foaming
within the mold.
The heat transfer from the furnace to the mold with the precursor material can
occur by three different mechanisms: conduction, convection, and radiation.
For the development of prototype manufacturing methods various furnaces
working with these three physical principles were tested. Basic calculations concerning heat capacity and heat transfer were carried out before testing the different
furnace systems. The heating rate in the foamable precursor material is the key
parameter, influenced by the foam part dimensions and foaming mold properties.
Considering the described intentions, the very first trials to make simple foam
parts were performed by immersion of the mold into an aluminum melt. To
achieve a high heating rate by intensive heat transfer by conduction, the molds
containing the precursor material were submerged into liquid aluminum that
was kept in a conventional holding furnace. The conductive heat transfer was
7.4.2.3
very high and a uniform temperature distribution could be achieved. This method
worked quite well and the aluminum-foam samples yielded good properties, but
chemical attack of the molds by the Al melt caused short lifetimes of the molds
and the mold attachments. Furthermore, the handling of the molds in liquid aluminum is dangerous and problematic.
High-convection furnaces (air-circulating furnaces) are used in industry for rapid
heating processes. A ventilator in the furnace effects very high heat transfer from
the heat source to the mold. The rapid heat transfer, caused by high flow velocity,
causes high heating rates, but the gas flow is influenced by the geometry of the
mold and that leads to uneven temperature distribution within the mold. This effect can be avoided by the installation of baffle plates, which must be adapted for
every new mold that is used in the furnace. Therefore air-circulating furnaces can
be employed in series production, where high numbers of aluminum-foam parts
have to be made with the same mold geometry, but they are not suitable for prototype manufacturing, where flexible and universal furnace systems are required.
For prototype manufacturing it is essential to cover a wide range of different
components. For this purpose a special chamber furnace with adequate heating capacity had to be developed. One example of a foaming furnace especially built for
foaming of big components, is the 100 kW Nabertherm Chamber furnace at LKR,
335
336
having internal dimensions 900 mm q 900 mm q 1800 mm and a modified control unit to keep the required temperature distribution tolerances within the chamber. This furnace is shown in Fig. 7.4-3.
Foaming Molds
In the PCF process the mold giving the aluminum-foam part its shape contains the
foamable precursor material during the heating cycle. The thermal energy that is
necessary to start the foaming of the precursor material has to be transmitted
through the mold. Consequently the thermal conductivity of the mold should be
as high as possible and the thermal capacity of the mold as low as possible. On
the other hand the molds must be stable enough to keep the required dimensional
tolerances of the net-shape foam part. The first molds for the PCF process were
made from common steel sheets. The very first aluminum-foam parts were very
simple, regular shaped pieces such as small plates, cubes, or cylinders and the development was focussed mainly on the foaming behavior of the precursor material
and the production of test samples. But when customers required 3D net-shape
prototypes, the mold became an important part of the prototype development.
The experience of foundry die makers was combined with the experience in foaming precursor material. Conventional casting dies are very heavy and bulky. A lot of
energy is wasted on heating the whole die during the foaming process. The precursor material would not be heated fast enough to achieve sound foam components.
An economic route is to use cast iron, which allows casting of thin walled foaming
dies, using wood or plastic patterns as well as the lost foam casting process. Depending on the molding sand used, the surface in the die may require machining.
7.4.2.4
Mould material
Advantages
Disadvantages
Steel
5 10 mm thick
easy machining
low material costs
heat conductivity
mold lifetime
distortion
Cast iron
heat conductivity
free-form shapes
High-temperature
tool steel
mould lifetime
low distortion
Copper
heat conductivity
Ceramics
shapeability
free-form shapes
no distortion
heat conductivity
mold lifetime
This method was used for foaming car structural parts, the largest prototypes so far
manufactured in net shape.
Besides structural steel and cast iron, high-temperature tool steel, copper, and
ceramics were also used as mold materials in a large number of prototype development projects. Table 7.4-1 shows an overview of different mold materials and
their properties.
To avoid sticking of the aluminum-foam product in the molds, the cavity of the
die was coated with common facing materials.
7.4.3
It is a long way from the first idea to the first prototype, and it is even further from
the prototype to a series product. Some of the product ideas had to be rejected because the aluminum-foam properties were over-estimated. The outstanding me-
Figure 7.4-4.
Aluminum-foam prototypes.
337
338
chanical properties of closed-cell metal foams are their energy absorbing potential
and specific stiffness. Figure 7.4-4 shows a couple of aluminum-foam samples
made for testing mechanical and technological properties of the material.
The feasible applications are spread in a wide range of various industrial fields.
The automotive and transportation industries are interested in metallic foams as
well as mechanical engineering, architecture, and building construction companies.
Automotive Applications
Owing to their particular properties, metal foams are considered to be an appropriate material for impact-energy absorbing components. Depending on foam type
(PCF or melt-based foam), alloy, and density, the energy absorbing behavior can
be influenced within a certain range. This property makes metal foams a candidate
7.4.3.1
types.
Energy-absorbing proto-
parts.
Structural
339
340
High specific structural parts such as intermediate panels or front panels are feasible applications. Figure 7.4-7 shows a couple of structural part prototypes including an intermediate panel [9] that was made for DaimlerChrysler in 1995.
The largest aluminum-foam prototype part made by LKR was a front panel that
was foamed by order of Opel in 1999. The dimensions of this panel, shown in
Fig. 7.4-8, are about 140 cm q 50 cm; it is a net-shape foamed in a cast-iron
mold [10].
Up until now sandwich sheets have had to be made in two process steps. A foam
plate is made and the face sheets are applied by adhesives. Manufacturing of sandwich parts and foam filled aluminum profiles by forming the foam between two
aluminum sheets is reported in Section 3.3.
Aluminum-foam filled steel profiles can also be used as energy absorbing
components in cars, trucks, trains, and other vehicles. Manufacturing of these
profiles is relatively easy because the steel profile can be used as a foaming mold.
In Fig. 7.4-10 a couple of foam-filled profiles and sandwich specimens are shown.
Aluminum or magnesium castings with permanent aluminum-foam cores
should meet the requirements of improved energy absorbing potential, higher
stiffness than hollow castings and good vibration damping behavior. The use of
foam cores can also improve the manufacturing process in a number of ways: no
sand cores have to be removed from the casting, complicated quasi-hollow shapes
can be cast, and improved solidification conditions can be achieved. Target applications for castings with foam cores, as shown in Fig. 7.4-11, are suspension and struc-
341
342
Aluminum foam
prototype part
Application field/Customer
Focus/property
Plates
construction, architecture
furniture industry
street furniture
weight, design
Cylinders, cubes
Tube
Continues
Aluminum foam
prototype part
Application field/Customer
Focus/property
automotive industry,
electrical appliance
camera casings
Structural parts
intermediate panels,
automotive structures
(DaimlerChrysler, Opel)
Structural parts
automotive structures
sunroofs
Stiffness, weight
Structural parts
weight
Sandwich structures
stiffness
weapons, stock
weight
Foam-filled profiles
crash behavior
Foam-filled profiles
damping properties
optical appearance
3-dimensional parts
sculptures, art
optical appearance
tural parts with improved energy absorbing properties, and wheels. Also mechanical
engineering components with good damping behavior and good castability [11].
In addition to the applications already described a large variety of other potential
technical applications of aluminum foam have been discussed. Table 7.4-2 contains an overview of feasible aluminum-foam applications.
Improbable Applications
As mentioned earlier, many of the possible applications of aluminum foams were
discussed with researchers and designers who had to solve very specific technical
problems, by use of suitable materials. People were fascinated by metal foams and
sometimes they over-estimated their potential. For instance, aluminum foam was
thought to provide the highest strength combined with lowest weight, which is
not the case. Table 7.4-3 shows a couple of improbable applications and ideas,
which could not be considered as applications for metal foams.
7.4.3.4
343
344
A very important task of foam developers and foam manufacturers must therefore be to give their customers and potential material users up-to-date and accurate
information about the material properties and convenient applications, but also its
limits. Otherwise they run the risk of losing interest in this material altogether.
Aluminum foam
prototype part
Application/Customer
Focus
Plates
safety, military
Plates
skateboard, snowboards
weight
Plates
milk tanks
weight
Aluminum foam
prototype part
Application/Customer
Focus
gear box
weight
Structural parts
musical instruments
acoustic properties
Plates, cubes
acoustic damping
Gas-permeable plates
weight
Gas-permeable cylinders
porosity, weight
Gas-permeable cylinders
heat exchangers
porosity
Foam cylinders
heat capacity
floats
density
Foam-filled cylinders
drums, rollers
textile machines
paper industry
Buttons
clothes
design
Precursor nails
References
345
346
7.5 Applying the Investment Methodology for Materials (IMM) to Aluminum Foams
7.5
IMM consists of three linked segments: viability analysis, market assessment, and
value capture (Fig. 7.5-1). A material is viable in an application if the balance between its technical and economic attributes is favorable. Assessing viability involves technical modeling of the application, cost modeling of manufacturing,
input from the market assessment, and value analysis. Market assessment deploys
techniques for identifying promising market applications and for forecasting future production volumes. Likelihood of value capture is assessed through an
Figure 7.5-1.
Aluminum foams have several unusual features, among them, exceptional energy absorbed per unit volume and, particularly, energy absorbed per unit weight [5].
Potential applications are those in which these functions are important; those
that also exploit secondary attributes of metal foams (such as flame resistance,
heat dissipation, noise reduction, water resistance, blast amelioration) are particularly attractive. The strongest opportunities for short-term take up of aluminum foam lie with substitution into existing markets, since for these the
design requirements are well known. Radically new designs taking best advantage of their properties are slowed by lack of designer awareness and confidence, and by current manufacturing difficulties with achieving high tolerances.
There are existing aerospace and automotive applications that require high energy absorption per unit volume. The large-scale production of aluminum foams
cannot yet meet the tolerances required for aerospace applications (other than
the expensive Duocel, which finds a market in defense and space vehicles); thus,
automotive applications are prioritized, specifically, the A-pillar designed for occupant safety, and the redesigned hood and front bumper regions to meet new pedestrian protection legislation.
7.5.3
Material Assessment
Technical Performance
The stress strain curve of aluminum foams (Fig. 7.5-2) approximates the idealized stress strain curve for energy absorption, suggesting their use in energy-absorbing applications. As a screening mechanism to eliminate prototyping and testing with uncompetitive materials, a range of aluminum foams
was compared with all other materials along combinations of technical parameters. IMM utilizes CES (Cambridge Engineering Selector) software for this
purpose [6]. Once a material is established as a non-dominated candidate for
an application, prototyping and testing are required to substantiate technical
viability. This technique is explained in detail in published books and papers
[7,8].
7.5.3.1
347
348
7.5 Applying the Investment Methodology for Materials (IMM) to Aluminum Foams
Figure 7.5-2.
(Duocell).
Stress strain curves of closed-cell Alporas and open-cell ERG metal foams
Cost of Production
Economic analysis aims to establish the manufacturing cost differential between a
component made with a novel process or material and the incumbent processes
and materials. To this purpose, technical cost models were constructed for three
processes for manufacturing aluminum foam.
7.5.3.2
x
x
x
The cost models themselves include sub-models to capture the effect of component
size on production rate, the die and equipment costs, allowance for scrap, die-life,
Figure 7.5-3.
and capital write-off. The output of the models (Fig. 7.5-3) shows the way in which
the cost of the manufactured component depends on production volume. Cost drivers can also be determined. In this instance the batch powder metallurgical process is the most economical option for annual production volumes of up to 20 000
parts [2,9].
By comparing the cheapest aluminum foam component (at the required production volume) with an incumbent material for any given application, the cost differential between a new and an incumbent material solution can be established. This
provides input into the value assessment.
Figure 7.5-4.
Selection of metal foams and polymer foams by volume and cost with s pl 3.5 MPa.
349
350
7.5 Applying the Investment Methodology for Materials (IMM) to Aluminum Foams
7.5.4
Market Forecast
A technical assessment indicates that metal foams are technically viable as energy
absorption elements within the A-pillar [8]. The next step is to examine the market
size and rate of take-up.
Figure 7.5-5.
cost/higher performance mode of substitution. Historical examples of these include aluminum alloy wheels and air bags. With aluminum wheels, a combination
of aesthetic qualities and lighter weight drove substitution in sporty platforms to a
50 % take-up after 18 years [13]. In the case of airbags, consumer safety concerns,
the viability of the innovation, and automotive regulation drove substitution much
more quickly [13]: an unlikely scenario in the case of metal foams in the A-pillar.
We conclude that, if all other conditions are favorable, the time to 50 % take-up
would be in the range 12 18 years.
7.5.5
Value Capture
Industry Structure
Porter's methodology for assessing industry attractiveness [12] directs attention to
the competitive threats and to buyer/supplier pressures that might reduce valuecapture by the innovating firm. In the present context we find that several companies are able to make metal foams and that none of them has tight control of the
7.5.5.1
351
352
7.5 Applying the Investment Methodology for Materials (IMM) to Aluminum Foams
Overall, medium-low
attractiveness
Figure 7.5-6.
Potential
Entrants
Suppliers
Competitive
Rivalry
Buyers
Substitutes
rights to the process, which is not difficult to reproduce. Their position is made
less attractive because they must fight both against competitors who are commercializing alternative processes for making metal foams (the lower end liquid aluminum foam process, for instance) and against substitutes for energy absorption in
automotive applications, such as polymeric foams, shaped aluminum sections, and
fiber-reinforced polymer composite sections. Additionally, the automotive companies exert very strong buyer power that will be difficult for the small producer to
counter (Fig. 7.5-6).
Appropriability of Profits
Teece's concept of appropriability [14] is useful in establishing the potential for
value capture. Here we seek to establish the appropriability position of a small
company seeking to sell metal foam automotive components. The intellectual
property (IP) position does not appear to be strong. Specialized assets exist with
the process, but most could be assembled by a die-casting competitor without excessive difficulty. There is a possibility for co-specialized assets, if certain methods
of automotive design propagate. Aluminum foam for energy absorbing automotive
applications is a revolutionary innovation, in that it overturns established technical
and production competencies, but does not overturn customer linkages nor require
a company to sell into different markets. New product cycle time is slow, giving a
longer period over which to appropriate value, and the structure of the automotive
industry allows for protection of IP. The conclusion is that the appropriability position is not strong, but might be classified as medium.
7.5.5.2
7.5.6
As the chances of value capture are relatively low and the payback period is relatively long, a small firm would be disadvantaged in commercializing this innovation unless as a joint venture with a metal supplier or automotive producer. A larger company might be interested in pursuing this opportunity if it was in a good
position to capture the value created. Alternatively, a government-sponsored initiative for pedestrian or occupant safety might subsidize the commercialization of
such an innovation.
Acknowledgements
We wish to acknowledge the support of the Korber Foundation, the Cambridge Canadian Trust, the UK EPSRC through the support of the Engineering Design Centre at Cambridge, and Granta Design, Cambridge, who developed the software
illustrated in this article.
References
353
354
7.5 Applying the Investment Methodology for Materials (IMM) to Aluminum Foams
9. E. M. A. Maine, M. F. Ashby, Cost estima- 12. M. E. Porter, Competitive Advantage: Creattion and the viability of metal foams in Proc. ing and Sustaining Superior Performance, Harvard Business School Press, Boston, MA
Int. Conf. Metal Foams and Porous Metal
1985.
Structures, MIT-Verlag, Bremen 1999, Vol. 1,
13. R. Grant, Contemporary Strategy Analysis,
p. 63 70.
Blackwell, Oxford 1998, p. 245.
10. W. J. Abernathy, K. B. Clark, Innovation:
mapping the winds of creative destruction 14. D. J. Teece, Profiting from technological
innovation: implications for integration,
Research Policy 1985, 14, 3 22.
collaboration, licensing and public policy.
11. E. M. A. Maine, Future of Polymers in
Research Policy, 15, p. 285305. (Elsevier)
Automotive Applications Master's thesis,
1986.
MIT 1997.
8
Strengths, Weaknesses, and Opportunities
H. P. Degischer
The research interest in cellular metals and the activities of manufacturers in various market sectors illustrate the appeal of these materials: they are attractive because of their unusual specific properties and appealing property profiles. Expectations for so-far unachievable metallic light-weight structures have created ideas for
the substitution of existing materials in weight-sensitive components such as
floating metals, light stiff panels for automotive and aerospace applications, crashabsorbing safety components, electromagnetic shielding, structural components
with low absorption for nuclear particle radiation, housings with high acoustic
damping, and decorative design objects. Many different markets have been
addressed, but how long will it take for a market breakthrough? Tables 8-1 8-3
resulted from a discussion at the end of the Symposium Metallschaume in
Vienna, 29 February 2000. The tables show the advantages of cellular metals and
the achievements of research as well as their disadvantages, and the remaining unanswered questions: i. e. cellular metals' strengths, weaknesses, and opportunities.
8.1
Processing
355
8.1 Processing
Table 8-1.
Type
Foamed metals
356
x
x
x
x
x
x
Type
Continues
Achievements in production
x
Industrial production
x
x
x
x
x
Ceramic inclusions;
no process industrially realized;
consistence and stability of substrates;
recycling requires separation from
substrate.
Various other techniques mostly in
laboratory status for research and
prototype production (for instance
COMBAL).
very low densities with the highest degree of regularity and reproducibility of cell
architecture. They seem to be only appropriate for small volume components for
high-tech and high-price applications.
Foaming of metals in the liquid state promises the most economic processing
route with the drawback of difficult control of cell architecture. Products of remarkable size and complexity have been made for structural applications. Although
foaming produces closed bubbles in the liquid, during solidification the cell
walls are usually damaged giving the components a certain permeability and negligible mechanical stiffening. Even the characteristic surface skin of foamed metal
parts can not be considered as defect-free and sufficiently protective. Processing of
closed-cell structures with enclosed hollow substrates or sintered hollow spheres is
still in the prototyping stage.
There are no defined specifications for cellular metals. They are essentially characterized by their processing method. The dependence of material properties on
the different techniques of production remain the subject of research and development. There is great potential in improving reproducibility by scaling-up production, allowing better control of processing parameters. Some successful examples
of industrial production are quoted in Table 8-1.
The chances of any of the cellular metals being used in components depends
mainly on the availability of appropriate bonding and joining techniques. Secondary processing like forming, machining, bonding, and surface treatment require
further technological developments, which have to be governed by the service requirements of each case. A significant advantage of cellular metals compared
with polymers is their recyclability and the potential of up-grading scrap metal,
which may gain increasing importance in the future.
357
8.2 Properties
8.2
Properties
The processing method for a cellular metal component defines the achievable
property profile in the first instance. The density achieved by a given production
route determines the magnitude of the properties. Physical principals govern
those property relations, which reveal the limits as well. The introduction of
pores into a metal cannot improve its mechanical properties even when related
to the specific mass. Tensile strength is limited by fracture mechanics due to inhomogeneous stress distributions within the cellular architecture containing many
crack-initiation sites. Properties depending nonlinearly on the dimension of the
part offer benefits for design application, for example specific stiffness for bending
of beams or plates.
The close relation between processing and properties is enhanced by the importance of the mesoscopic 3D-structure formed by the metal matrix with its specific
microstructure. The scatter of material properties is assumed to be caused by irregularities in the cellular architecture of real products. Some of these can be tolerated, but there is still the ambiguity of extremely poor property values not yet correlated with structural parameters. Sensitivity studies are required to relate the influence of meso- and microscopic defects to the properties. Consequently the reliability of the properties needs verification by testing series of products.
The properties depend on the variations in cellular architecture and their uniformity. The actual achievements are contrasted in Table 8-2 with the deficiencies still
existing for most of the available cellular metals. The average density does not
seem sufficient for the characterization of a cellular component. The properties
of cellular materials are not only determined by the matrix metal, but essentially
Table 8-2.
Topic
Cellular structure
358
x
x
x
Characterization deficiencies
x
Topic
Continues
Available property data
x
Mechanical properties
Characterization deficiencies
x
x
x
x
x
Physical properties
x
x
Environmental properties
x
x
x
x
359
360
by the local material distribution: the architecture of the cell structure. X-ray computed tomography offers the possibility of 3D characterization, but requires significant effort. Nondestructive testing is one of the main activities of the research necessary to correlate properties with cellular architecture. Furthermore, the properties of the matrix metal itself cannot be simply transferred from bulk material, but
need to be investigated for each processing technique. In the case of age-hardening
metals, the influence of thermal treatments during production has been indicated,
but may offer chances for further material development.
Table 8.2 lists the most appealing properties, especially attractive in relation with
the low specific mass. The potential of cellular metals depends on combinations of
these specific properties according to particular service requirements. The mechanical properties for structural applications have to be defined specifically for these
cellular structures. Elastic properties, for example, have to be considered as the
stiffness properties of the architecture of the metal; similarly plastic deformation
and damage mechanisms. Complications arise from the internally multi-axial
load transfer, which require special attention regarding the test conditions providing material data. Both test conditions and the significance of the results have to be
defined properly for cellular metals.
Life-time predictions require further research to complete the achieved results
for exposure conditions simulating long-term service. Damage mechanisms are revealed qualitatively and first attempts of quantification need verification in use. The
physical properties deserve considerable attention to underline the perspectives of
multi-functional application. They have to be determined quantitatively individually for each type of cellular metal, based on the principally existing potential.
There are essential advantages of cellular metals in comparison with polymers regarding environmental properties. Of course, the environmental service conditions
have to be considered for specific cases. The high fire resistance and the space
compatibility have not yet been exploited for aerospace applications.
8.3
The technical advantages of cellular metals cannot be fully exploited just by substitution of materials, but require material-appropriate engineering solutions and design concepts. The simulation of the performance of components relies on the material laws and constitutive equations appropriate to cellular metals and their loading conditions. The actual situation in modeling of cellular metals and in designing components thereof is summarized in Table 8-3. Material models for cellular
metals are required, which have to reflect the 3D behavior of the highly porous,
compressible material. Theoretical studies help to identify the important parameters. The transfer of the theoretical results to real architectures requires experimental calibration of the material laws. As stated before, even micromechanical
computations require the appropriate metal matrix properties, which are very difficult to determine in the processing condition of cells. Reasonable approximations
products.
Topic
Actual achievements
x
Modeling
x
x
361
Topic
Continues
Actual achievements
x
Application
Market
362
x
x
x
x
x
x
x
x
based on averaged or characteristic structures are available. The hierarchical approach of macro-, meso-, and micromechanical simulations according to the
local sensitivity of a structural feature in relation to the load case of the component
seems very promising.
The close interaction of development of models with experimental determination
of properties is leading to useful interpretations of the general material behavior.
The access to the material data required for input to the different models is a specific topic of research. Finite-element calculations have to take care of the appropriate choice of mesh geometry owing to the generic heterogeneous structure dealt
with, whether for the micromechanical considerations or for the mesomechanical
363
Index
384
481
482
Index
copA 368
copB 368
cop operon of Enterococcus hirae 368
copY 368
copZ 368
CopA, in copper uptake 369
b
CopB, in copper excretion 369
b-barrel porins 6
CopB copper ATPase, of Enterococcus
Bacillus stearothermophilus 49
hirae 363
Bacillus subtilis, CitM protein 404
copper ATPases
saturable Mn permease 330
of Helicobacter pylori 373
Bacillus thuringiensis Cry toxins 223
of Listeria monocytogenes 373
bacterial copper transport 361 ff
copper chaperones 366
bacterial cytochrome c oxidases 436
copper excretion, CopB 369
bacterial iron transport 289 ff
copper ion transport, mitochondrial- 419 ff
bacterial zinc transport 313 ff
copper resistance
binding protein-dependent Zn2 uptake
in Escherichia coli 371
in gram-negative bacteria 320
plasmid-encoded- 374
in gram-positive bacteria 316 ff
copper transport 361 ff
Bradyrhizobium japonicum (HupN) 413
copper transport system, in S. cerevisiae 426
Bsd2p and manganese ions 453
copper uptake, CopA 369
coprogen transporter 472
c
Ca2 -ATPase, of sarcoplasmic reticulum 364 CopY repressor protein 370
CorA
Ca(II) ATPase 428
of Methanococcus jannaschii 353
Ca-ATPase calcium pump, structure of 428
of Salmonella typhimurium 352 ff
Caloramator fervidus 49
CorA magnesium transporter 351 ff
Candida albicans peptide transport gene
Corynebacterium glutamicum 53
(Ca Ptr2) 155
COS16 457
carbon monoxide dehydrogenases [Ni] 399
Cox polypeptides 425
catalysis within the F1 complex 27 ff
Cox1 polypeptide 424
cation diffusion facilitator 315
Cox2 subunit 424
Ccc1p: a manganese homeostasis protein
Cox11 sequences 434
456
Cox17 metallochaperone 429
Ccc2 P-type ATPase transporter 427
Cox17 sequences 430
CCC2, intracellular transporter of copper
CPx motif 364
449
CPx-type ATPases 362
Ccc2p 456
sequence motifs 365
CDC1 gene product 456
crystal structures of DtxR 305
cell surface permease FTR1 449
CTR1, cell surface copper permease 449
channel formation steps 212 ff
Ctr1/Ctr3 system 426
channel forming colicins 222
Cu metallation, of cytochrome c
channel forming protein toxins 209 ff
oxidase 431 ff
a-toxin 217
Cu metallation, of mitochondrial cytochrome c
aerolysin 215 ff
oxidase 441
channel forming proteins 209 ff
Cu-specific ATPase pumps 428
classification of 211
CxxC motifs 366
cholesterol-dependent toxins 219
cytochrome c oxidase, structure of 424
CitM protein of Bacillus subtilis 404
cytochrome c oxidase
citrate fermentation 51
Cu metallation of 431 ff
classes of transporters 4 ff
Cu metallation of mitochondrial- 441
Clostridium fervidus 49
mitochondrial- 422 ff
Cnr (cobalt nickel resistance) 406
cytochrome c oxidases, bacterial- 436
coenzyme F430 399
Czc (cobalt zinc cadmium resistance) 406
colicins 222
Index
des(diserylglycyl)ferrirhodin 470
dipeptide permease (Dpp) 144
diphtheria toxin 304
DMT1 329
Dpp-type 159
DppA 89
DTAP domain, in ArsA 386
DtxR, crystal structures 305
DtxR/IdeR family 338
EI 118
enantio-ferrichrome 470
enantio-rhizoferrin 468
ENB1 transporter 472
energy transduction, ATP synthase 31 ff
enteric bacteria, peptide transport in 143 ff
Enterococcus hirae cop operon 368
Enterococcus hirae CopB copper ATPase
363
Escherichia coli HlyA 221
aquaporin Z 250, 255
ars operon 378, 391
ABC transporters 80 ff
ArsA 388
ATP synthase 24 ff, 36 ff
ATP transporter binding proteins 90
copper transport in 371
FhuABCD activities 292
FhuD protein 297
LamB 240
MalK 105
melibiose transporter 54
Na/melibiose transporter 58
Na/proline transporter 55, 59 ff, 65 ff
Na/substrate transport in 52
Nik system 407
OmpF 234
peptide transport 143 ff
Tat pathway in 174
TolC trimer 242
extracellular loops
in FepA 270 ff
in FhuA 270 ff
F1 subunits 25 ff
F1/FO interface 31 ff
F1FO ATP synthase 23 ff
model of energy transduction 39 ff
FO subunits 33 ff
proton translocation pathway 42
F-ATPases 23 ff
F-type ATPases 23 ff
families of transporters 1 ff, 5 ff
Fe3 -citrate 306
Fe3 -dicitrate 300
Fe3 regulatory mechanism 306
Fe3 -siderophore transport 291 ff
Fe-enterobactin transporter 473
FecI and FecR proteins 306
feo genes 304
FepA 264 ff
b-barrel 265 ff
conserved residues 284
extracellular loops 270 ff
N-terminal domain 267 ff
ribbon diagram 264
structure determination of 275
FepA mutants 285
Ferric Uptake Repressor (Fur) 337
ferric citrate transport system 306
ferrichrome 470
ferrichrome A 470
ferrichrome transporter 471
ferrichrysin 470
ferricrocin 470
ferrirhodin 470
ferrirubin 470
ferrous iron transport system 304
FET3 448
FET4 450
FET5 450
FhuA 264 ff
barrel structure 266
conserved residues 284
extracellular loops 270 ff
N-terminal domain 267 ff
structures with ligand 272 ff
FhuA protein 265, 292
FhuABCD activities, of E. coli 292
FhuB transport protein 298
FhuC, ATP hydrolysis by 298
FhuD 89, 295 ff
crystal structure 297
Fibrobacter succinogenes 50
FRE genes, of Saccharomyces cerevisiae
448
FRE reductases, in siderophore
transport 474 ff
FRE1 447
FRE2 447
FRE3 448
FRE4 448
FTR1 448
fungal ornithine-N5 -oxygenase 466
fungi, siderophore transport in 463 ff
483
484
Index
Klebsiella pneumoniae
citrate fermentation in 51
melibiose transporter 54
Na/citrate transporter 56, 59, 61, 69
H cycle 52
Haemophilus influenzae
ATP transporter binding proteins 90
heme transport systems 301
hFBP 297
Halobacteriales 50
Halobacterium 50
HbpA 89
heavy metal ATPases 362
heavy metals, transport of 452
Helicobacter pylori, nickel/cobalt
transporters 409
Helicobacter pylori copper ATPases 373
heme, bacterial use of 300
heme transport systems 301 ff
hexahydroferrirhodin 470
HgAtx1 complex, structure of 427
high-affinity nickel uptake systems 406
HisJ 89
HoxN 410
hpCopA 373
HPr 118
hydrogenases [NiFe] 399
hydroxamate reductases 476
IIA 118
IIAGlc subunit 120
regulatory role 131
IIB 118
IIC 118
IICBGlc mutants 124 ff
listing of 125 ff
IICBGlc subunit 121 ff
regulatory role 132
topology of 122
insecticidal crystal (Cry) proteins 223
integral transmembrane domains (TMDs)
ABC transporters 91 ff
sequence 96
ion selectivity of porins 233
Index
manganese transport
Saccharomyces cerevisiae genes involved
in 458
in Saccharomyces cerevisiae 452 ff
in bacteria 330 ff
manganese transport ATPase, Pmr1p 455
manganese transporters, in yeast 447 ff
manganese uptake, in bacteria 325
MBP 89
MBP (MalE) 89
melibiose transporter of E. coli (EcMelB)
54, 62 ff
melibiose transporter of K. pneumoniae
(KnMelB) 54, 62
membrane topology of CPx-type ATPases 363
metal ABC transporters 316
metalloenzymes, nickel as cofactor 398 ff
Methanobacterium thermoautotrophicum 49
Methanococcus jannaschii 49, 353
MFS-type nickel exporter (NrsD) 406
Mg2 transporter families: MgtE, CorA,
and MgtA/B 347
MgtA/MgtB Mg2 transporters 355 ff
MgtC protein 357
MgtE magnesium transporters 350
microbial nickel transport 397 ff
microbial siderophores 465
MIP-like channel genes 252
MIP-related aquaporins 248
MIP-related sequences, phylogenetic
tree 252
mitochondrial copper ion transport
419 ff
mitochondrial cytochrome c oxidase
422 ff
MMT1 451
MMT2 451
Mn(II) ABC transporters 328
Mn(II) as a Fur co-repressor 338
Mn(II) primary transporter 328
Mn(II) permeases 328
Mn(II) transport in S. aureus 330
MntABC uptake system 333
MntH 329
MntH proteins 336
ModA 89
molecular recognition templates
(MRT) 156 ff
molecular recognition templates (MRTs),
optimal features of 159
MRP1 (multidrug resistance-associated
protein) 393
multi-copper oxidase 448
Mycobacterium tuberculosis (MtNicT) 413
N-terminal domain
of FepA 267 ff
of FhuA 267 ff
Na cycle 52
Na/citrate transporter (CitS) of
K. pneumoniae 56, 59, 61, 69
Na/H antiporter 52
Na/K ATPase 23
Na/melibiose transporter of E. coli 58
Na/proline transporter 53
Na/proline transporter (PutP) of E. coli
59 ff, 55, 65 ff
Na/substrate transport in Escherichia coli
52
Na/substrate transport systems 48 ff
Natural Resistance-Associated Macrophage
Protein (Nramp1) 329
Ncc (nickel cobalt cadmium resistance) 406
NhlF 410
Nic1p, of Schizosaccharomyces pombe 412
nickel, as a cofactor of metalloenzymes
398 ff
nickel/cobalt transporter family 408 ff
hydropathy profile alignment 409
in Bradyrhizobium japonicum 412
nickel homeostasis 403 ff
nickel resistance 401
nickel toxicity 401
nickel transport 397 ff
nickel transporter (UreH, UreI)
of thermophilic Bacillus sp. 412
nickel transporters, ABC-type- 407
nickel uptake, high-affinity systems
406
nickel/cobalt transporters, metabolic
functions 410
NiCoT in Bradyrhizobium japonicum,
nickel/cobalt transporter in 412
Nik system, Escherichia coli 407
Nik-related transporters 408
nikABCDE operon 407
nikR gene 407
nitrous oxide reductase (N2OR),
in P. stutzeri 439
NixA 410
(Nramp)/Divalent Metal Transporter
(DMT) 329
Nramp family 452
Nramp1 protein 339
Nramp2 329
nucleotide binding sites, in ArsA
381 ff
485
486
Index
Index
487
488
Index
Vibrio alginolyticus 50
Vibrio parahaemolyticus 50
virulence factors 304
vitamin B12, binding of 294
yeast
ACR genes 393
AFT1 in 466
iron transporters 447 ff
manganese transporters 447 ff
yeast oxidase complex 422
yeast vacuole and manganese 457
Yersinia pestis YfeABCD system 334
YfeABCD system, of Yersinia pestis 334